Applying a machine-based learning algorithm to images captured with CT colonography makes it possible for radiologists to differentiate between benign and pre-malignant colorectal polyps.

In industrialized countries across the globe, colorectal cancer ranks among the top three most common causes of cancer-related deaths among both men and women. In a study published Feb. 23 in 

a team of investigators from Germany showed how using radiomics can expand CT colonography beyond simply detecting polyps to pinpointing which ones are likely to progress to being cancerous.

“CT colonography does not enable a definite differentiation between benign and premalignant polyps, crucial for individuals risk stratification and therapy guidance,” said the team led by lead study author Sergio Grosu, M.D., a radiologist from University Hospital at Ludwig Maximilian University in Munich. “In this proof-of-concept study, machine learning-assisted CT colonography analysis allowed for differentiation of benign and premalignant colorectal polyps at high diagnostic performance associated with the histopathologic reference standard.”

A, Optical colonoscopy and, B–D, CT colonography of a 9-mm polyp (arrow) in the descending colon of a 78-year-old woman. B, Virtual fly-through three-dimensional reconstructions were used for exact polyp localization. C, Manual polyp segmentation was performed in multiplanar two-dimensional CT colonography images. D, CT colonography images were preprocessed for image feature extraction by application of a dedicated filter.

For nearly 20 years, CT colonography has gained popularity as a non-invasive screening method for colorectal cancer. It performs equally as well as optical colonoscopy (OC) in visualizing polyps that are at least 6 mm large in asymptomatic patients, and it is also provides better images of complex anatomical cases for which colonoscopy is frequently ineffective.

But, Grosu’s team investigated whether it was possible to push the screening method’s diagnostic value even further to potentially allow for pre-cancerous polyp identification.

To reach this goal, the researchers created a machine learning algorithm that uses radiomics to extract quantitative image features to predict whether the polyps would further develop into cancer. They applied this algorithm to CT colonography images taken in a group of asymptomatic patients who were at average colorectal cancer risk. They, then, trained the algorithm on 169 CT segmentations from 107 colorectal polyps from 63 patients and tested it on 118 segmentations from 77 polyps from 59 patients.

Top: Axial CT colonography images show representative colorectal polyps (arrow) in the training set. Bottom: Corresponding histopathologic work-up. (Hematoxylin-eosin staining; original magnification 320.) A, An 8-mm hyperplastic polyp in the ascending colon of a 54-year-old woman with hyperplastic epithelia. B, An 8-mm tubular adenoma in the sigmoid colon of a 68-year-old man with tubular growth pattern and elongated nuclei. C, An 11-mm tubulovillous adenoma in the rectum of a 73-year-old man with tubulovillous growth pattern and elongated nuclei.

Using random forest analysis, the team determined that their radiomics-based machine learning method allowed for non-invasive differentiation of benign and pre-malignant colorectal polyps with 82-percent sensitivity and 85-percent specificity. The area under the curve (AUC) was excellent, they said, at 0.91.

When the team examined the method’s performance based on the size of the polyps detected, they found that for polyps between 6 mm and 9 mm, the AUC was 0.87 – it was 0.90 for polyps that were at least 10 mm in size. The team also found that the most important image feature for decision making was quantifying first-order gray level histogram statistics.

These results fall in line with existing studies, the team said, but their findings add to the knowledge base around the use of CT colonography for polyp identification because they examined polyps that were smaller than 8 mm. In addition, their algorithm was trained on a balanced training set, and it was validated with an external multi-center test set.

Overall, Grosu said, incorporating a machine learning-derived algorithm can make CT colonography an even better colorectal cancer screening tool.

“Adding machine learning-assisted image analysis to conventional, radiological image reading could further improve the clinical significant of CT colonography-based colorectal cancer screening by allowing for a more precise selection of patients eligible for subsequent polypectomy,” he said. “Furthermore, machine learning-assisted CT colonography image analysis potentially provides the gastroenterologist with a road map of pre-malignant polyps eligible for OC-guided resection, particularly in patients with a high burden of colorectal polyps, such as in familial adenomatous polyposis.”

Still, he said, additional larger studies are needed to further validate the findings and refine the algorithm, making it better suited for the clinical environment.

Articles

Implementing a radiomics-based machine learning algorithm allows CT colonography to differentiate between benign and pre-cancerous polyps with high sensitivity and specificity.

CT Colonography Identifies Pre-Malignant Polyps with Machine-Based Learning Algorithm

Case submitted by Mohammad Osama Yonso, M.D., radiologist, to

For more coverage based on industry expert insights and research, subscribe to the Diagnostic Imaging e-Newsletter 

 Asymptomatic 35-year-old male presents with respiratory symptoms to rule out pneumonia. No focal airspace opacities are seen to suggest pneumonia, but there is a congenital variant of right-sided aortic arch and knob visualized. No aneurysm or other vascular variant are identified. No situs inversus is seen.

 Right-sided aortic arch is a variant characterized by the aortic arch with course to right of the airway. Various patterns can be seen with patterns of right-sided aortic arch associated with mirror image branching or right-sided aortic arch associated with with aberrant left subclavian artery (ALSA). 

On chest X-ray, look for the absence of left-sided aortic contour. The trachea tends to bow towards the left slightly. Right-sided descending aorta is also visualized adjacent or overlying the spine. Finding can be mistaken as a mass in the right paratracheal region.

What is your diagnosis on this patient who got a CXR to rule out pneumonia?






Image IQ Quiz: Asymptomatic Patient Presents with Respiratory Symptoms

Even though it is rare, COVID-19-positive patients can develop leukoencephalopathy – white brain matter diseases – according to findings captured with brain MRI.

a team of investigators from the University of Pennsylvania (UPenn) stressed that radiologists must keep their eyes open for this possibility to appear on images.

“Increasingly, effects of COVID-19 on the brain are being reported,” said the team led by Colbey Freeman, M.D., a fourth-year radiology resident with the UPenn, “including acute necrotizing encephalopathy, infarcts, micro-hemorrhage, acute disseminated encephalomyelitis, and leukoencephalopathy.”

Currently, the industry does not have any criteria for defining COVID-19 disseminated leukoencephalopathy (CRDL). To help with potential diagnosis, Freeman’s team investigated whether findings existed that could facilitate identification.

63-year-old woman with history of hypertension and type 2 diabetes mellitus who presented with coronavirus disease (COVID-19) pneumonia complicated by acute respiratory distress syndrome.

A, Axial FLAIR MR image shows areas of T2 prolongation in posterior inferior cerebellar hemispheres (arrows).
B, Contrast-enhanced T1-weighted image shows enhancement of areas with T2 prolongation (arrows).

Courtesy: American Journal of Roentgenology

For their study, they examined 2,820 COVID-19-positive patients who were admitted to the hospital between March 2020 and June 2020. Brain MRI was performed on 59 patients (2.1 percent). Any scans that revealed white matter abnormalities that were not typically associated with other causes, such as hypoxic injury or chronic small vessel ischemic disease, that were picked up in the bilateral middle cerebellar peduncles or corpus callosum were considered suspicious for COVID-19, the team said.

Based on their findings from T1-weighed and T2-weighted FLAIR images, the team identified six patients (10.2 percent) who were suggestive of CRDL.

“[These images were] characterized by extensive confluent or multifocal white matter lesions (with characteristics and locations atypical for other causes), microhemorrhages, diffusion restriction, and enhancement,” the team said.

In addition, hypertension and type 2 diabetes were common in these six patients, affecting four (66.7 percent) and three (50 percent), respectively.

Alongside the CRDL-suggestive findings, the team also identified other neurological conditions:

Three (5.1 percent) with known multiple sclerosis white matter lesions.

23 (39 percent) had small vessel ischemic disease white matter lesions.

Four (6.8 percent) had subacute infarcts.

One (1.7 percent) had abnormal basal ganglia signal from hypoxemia.

Two (3.4 percent) had micro-hemorrhage associated with chronic infarcts.

Two (3.4 percent) had acute or subacute infarct-associated micro-hemorrhages.

The team did note that their study was limited because current CRDL criteria do not exist. Additionally, these findings could overlap with findings from other conditions, potentially over- or under-indicating CRDL cases. Still, they maintained that radiologists must stay alert to the potential presentation of this condition.

“CRDL is rare, limiting our ability to assess the full array of possible findings,” the team concluded. “Nevertheless, leukoencephalopathy represent an uncommon, but important, differential consideration in patients with COVID-19 with neurological manifestations.”

Images can show indications of rare condition that can affect white brain matter in patients who test positive for the virus.



Brain MRI Reveals Leukoencephalopathy Vulnerability in COVID-19 Patients

In this week’s preview, here are some highlights of what you can expect to see coming soon:

CT conolography has been growing in popularity for several years, largely due to the benefits for patient comfort. New research is now showing advancements and benefits of using machine learning with this method of screening for colon cancer. Look for coverage in the coming days of a study being published later this week.

Pinpointing the correct treatment strategy for patients with acute ischemic stroke can be difficult. But, incorporating MRI with X-ray angiography can make the process easier, leading to improved patient outcomes. Findings published in 

 recently shows that this combination works well to guide providers in deciding on the best therapy. Keep your eyes open for this story later in the week.

For more coverage acute ischemic stroke, 

February is American Heart Month, and state-of-the art, cutting edge cardiovascular imaging is critical to helping providers offer the best heart care possible. This week, 

 spoke with industry experts about what is on the horizon with cardiovascular imaging. What does the future look like, and what does this mean for both patient and provider? Look for the story later in the week.

For additional coverage on cardiovascular imaging

Here's what to expect this week on Diagnostic Imaging.

Diagnostic Imaging's A Look Ahead: Feb. 22, 2021, to Feb. 26, 2021

Over the almost-decade (yikes!) I’ve been writing this column, I’ve made reference to some of my nerdly sources of entertainment: Comic books, sci-fi/fantasy TV shows, and movies, etc. I suspect a decent chunk of my reading-audience shares such interests. Indeed, enough of that stuff has gone mainstream that knowing/liking it doesn’t carry much of a nerdish stigma anymore.

So much so, I’d be surprised if anybody hadn’t encountered the concept of alternate realities. Marvel had an entire comic-book series, and I think a forthcoming streaming show dedicated to it (“What If…?”). DC’s version was called “Elseworlds.”

It’s a handy enough plot device, especially back when there was more of a perceived need for fiction to serve as a vehicle for moral/societal commentary. Good ol’ 

 had its “Mirror Universe,” not only to entertain viewers with evil versions of Captain Kirk and Mr. Spock, but also to highlight the goodness of the regular characters by comparison.

It even works in reverse: A comic from Marvel just had one of their best villains visit an alternate universe in which his counterpart was a good guy, instrumental in making the world utopian. (The bad guy had intended to reform and do the same back home, but ultimately took exception to something his alternate self said and murdered the guy…then destroyed the alternate universe for good measure.)

So I found myself thinking about what it would be like to encounter a radiologist from a world in which diagnostic imaging, from our perspective, got everything right that we have wrong.

The our-world rad would be wowed, and more than a little envious, to hear about how good the other-world rads had it: They always got relevant, accurate clinical histories for the studies they read. Referrers knew and utilized appropriate modalities and protocols for their clinical questions—or, when they didn’t know what to order, they deferred to their rads for best practices. Support-staff were routinely competent and eager to help (in other words, supportive).

Other-world software worked reliably (“Why wouldn’t it? That’s what it’s there for,” the other-world rad would ask us with genuine puzzlement). When rads or other “stakeholders” identified potential improvements, they were implemented, even invented. Workflow protocols were structured to optimize the performance (and satisfaction) of those doing the work.

Part of the whole alternate-reality gimmick is to see how the otherworldly characters react to how we do things. In this case, with pitying disgust and/or horror. How little control the rads of our world have over their field! They’re held accountable for things they can’t do anything about, and they’re on a perpetual slide towards even less authority. Also less compensation and ever-increasing demands on them. Meanwhile, in the other world, they’re respected professionals, exercising responsible control over their own destinies.

I half-baked a fictional scenario to showcase this. A this-world radiologist cobbled together the prototype of a new-fangled quantum-imaging scanner, but during a glitchy test-run, the gantry opened a portal to the paradisical alternate reality. On the other side, a similarly-inventive rad quizzically peered through from his prototype machine. Their my-world/your-world discussion ensued.

Before I determined there was no way I was going to fit such a story into the space I use here—I have a hard enough time confining myself as it is—it occurred to me that the story would need a “twist” to be really interesting. Thus, the quantum-gateway would initially showcase the paradisical other world, but later on would hiccup and add a third reality—one far worse off than our own. On the other side of that window, there’d be a beaten-down shred of a rad for whom even our imperfect reality would be the answer to a prayer. Because, let us not forget, no matter how bad one may think things are, they could 

The plot could have gone any number of ways from there, but a major theme would be the imperfect and hellish-world rads wishing and wondering if there were any realistic ways their societies could take steps to more resemble the paradisical one. And the paradise and imperfect-world rads would consider their less-fortunate counterparts as cautionary tales; what might they do to avoid heading down similar dystopian paths in the future?

My favored notion for an ending had the hellish-world rad concluding that his society was too far gone; no matter what he brought back from this conversation, it would improve nothing. Instead, having seen how things could be so much better, his subsequent life would be that much more torturous. Thus, before the unstable quantum-portal collapsed, the worst-off of the rads climbed through it to live out the rest of his life in the radiology-paradise world.

The imperfect-world rad was invited to join them, but declined: Having seen how good (and bad) things could get, he chose to stay put. He had faith that he could make a difference in his native reality, making things better…or at least preventing them from getting worse.

Follow Editorial Board member Eric Postal, M.D., on Twitter: 

Radiology, in an imperfect world, can always be improved.

Alternate Radiologists

3D mammography is not equally accessible to all women. Women who come from racial or ethnic minority groups, women who are less educated, and women who have lower incomes are less likely to undergo the exam that is most likely to detect an existing breast cancer without raising a false red flag.

In a five-state study, including nearly 100 facilities, a multi-institutional team showed that less than half of women in these groups have access to this type of breast cancer screening.

“This study was about whether adoption of this technology is equitable. We’re showing that it has not been, even though it has been FDA-approved for a decade now,” said lead study author Christoph Lee, M.D., MS, MBA, professor of radiology from the University of Washington School of Medicine. “Black and Hispanic women, and less-educated and lower-income women have not been able to obtain 3D mammography as easily as white, well-educated, and higher-income women.”

Lee’s team published their findings on Feb. 19 in 

Digital breast tomosynthesis (DBT) has been a viable alternative to digital mammography since it received approval from the U.S. Food & Drug Administration in February 2011. In the intervening years, a plethora of studies have shown that DBT is actually more accurate than digital mammography and does a better job at detecting breast cancer with fewer false positives.

Digital Breast Tomosynthesis on Track to Replace Digital Mammography

For this study, the team examined 2.3 billion breast screening exams that were pulled from the Breast Cancer Surveillance Consortium. Patients, who were screened between January 2011 and December 2017, came from 92 imaging facilities across five states. With this volume, the team said, it is the largest study, to date, to analyze the access women have to DBT throughout the United States. In addition to looking at whether access was equitable among all women, the team also assessed whether facilities offered DBT at the time a patient was screened and compared DBT and mammography use at each facility across patient populations.

Their findings showed that DBT use and access did change over time. At the start of the study in 2011, only 3 percent of women included had access to DBT. By 2017, more facilities were using the technology, and 82 percent of women had access – but, it was not equal across all groups.

“Given the large research sample and our longitudinal data collection, we were able to evaluate use by minority and traditionally underserved populations,” said senior author Diana Miglioretti, Ph.D., professor and division chief of biostatistics from the University of California Davis. “Unfortunately, we were not surprised to find that these traditionally underserved populations were less likely to attend facilities that offered 3D mammography, and even when they did, they less likely to receive a 3D mammogram.”

According to their analysis – when both 2D and 3D mammograms were available – DBT was utilized by fewer than 50 percent of women who were in racial and ethnic minority groups, as well as those who were less educated and those who were of lower income.

Specifically, a DBT scan was provided to:

37 percent of African American women, 43 percent of Asian-American women, and 44 of Hispanic women versus 53 percent of white women

41 percent of women with less than a high school education compared to 50 percent of women who had earned a college degree

44 percent of women living in zip codes that had the lowest 25 percent of median household incomes versus 51 percent of women who lived in zip codes with the highest 25 percent of median household incomes

These results are critical, Lee said, because women with less access to 3D screening are already traditionally underserved, putting them at a greater risk of breast cancer morbidity and mortality.

Although addressing whether the institutional barriers to care, such as structural racism or cost, contributed to decreased access to newer technologies was outside the scope of the study, the team acknowledged that these obstacles are real and likely played into the inequity.

For example, Lee said, because DBT generates digital slices of breast tissue, acquisition and interpretation time are longer, and the cost is higher. While some states, including Washington, have laws in place that prevent charging patients more for DBT, most states do not. Consequently, depending on the type of insurance a patient has, she may have to pay high out-of-pocket costs for a 3D mammogram.

Additionally, a woman’s education could impact how much she knows about DBT mammography and the benefits it can provide. Women who have higher education levels could be more likely to explore healthcare options, to research the possible benefits of 3D mammography, or to search for a facility that offers the technology. They could also be more likely to directly ask for the service.

The physical availability of the technology also played a role. While approximately 40 percent of all certified mammography units actually have DBT capability, more than two-thirds of breast imaging facilities offer DBT in at least one of their units, according to FDA data. In addition, urban or rural location was less important than anticipated, Lee said.

In such situations, he said, a rural site could have one digital mammography machine that would provide access to all patients once switched to DBT. However, urban facilities might have more money to purchase machines, but if it can only afford one 3D-capable unit – at a cost of approximately $750,000 a pop – they will still have to direct some patients to 2D machines. Consequently, some sub-groups of women are not being screened with 3D mammography even when the option technically exists at their imaging facility.

“We’re going in the wrong direction. You have a lot more women in certain subpopulations benefitting from new technologies and other subpopulations not,” he said. “Existing disparities in breast cancer screening outcomes could widen unless these factors are addressed.”

Women with less education, with lower income, or from racial and ethnic minority groups have less access to 3D mammography and the potential benefits it provides. 

Race, Lower Income, Less Education Reduce DBT Access

Welcome to Diagnostic Imaging’s Weekly Scan. I’m senior editor, Whitney Palmer.

Before we get to our featured interview with a multi-institutional panel about efforts to expand radiology education and the use of radiology technologies globally, here are the top stories of the week.

The “sweet spot” recall rate for screening mammography is actually lower than the currently accepted level, according to new research. Rather than the 12 percent-to-14 percent recognized, new research published in the 

shows that the appropriate recall rate for 2D digital mammography and 3D tomosynthesis is 7 percent-to-9 percent. Following this new benchmark could potentially reduce the number of women who are called back for additional diagnostic imaging. To identify the optimal recall rate, the team examined nearly 1.1 million 2D and 3D exams from women between the ages of 40 and 79 from January 2005 until December 2017. Based on their evaluation, they identified an overall mean recall rate for mammography of 10.7 percent and a cancer detection rate of 4 per 1,000 studies. The biopsy recommendation rate was 1.6 percent. When they drilled down, though, they discovered that 2D analog exams had the highest recall rate at 11.2 percent, followed by 2D digital exams at 10.8 percent and 3D tomosynthesis at 9.4 percent. 3D tomosynthesis also had the highest mean biopsy recommendation rate at 1.93 percent, followed by 2D digital and 2D analog exams at 1.55 percent and 1.47 percent, respectively. These results, they said, showed that a recall rate of 7 percent-to-9 percent maximized cancer detection and minimized unnecessary biopsy.

Using MRI, instead of CT, for follow-up in men who have been treated for early-stage testicular cancer reduces their exposure to additional radiation and could lower their risk of developing future cancers. Currently, treatment for these men involves removing the affected testicle and following the patient with CT for five years. However, new research presented at the 

 conference this month shows that MRI performs just as well at detecting any cancer relapse without the extra exposure. Investigators reached this conclusion by testing three follow-up protocols – the standard seven CT scans, three CT scans, and the same regimens with MRI – with 669 men with stage-one testicular cancer who had their affected testicle removed. Only 12 percent of the men had a relapse, 10 of which were advanced disease at detection. Of the cancers, nine were detected in the three-scan group, and only one was picked up in the seven-scan group, indicating that any additional CT scans after three years are unnecessary. CT detected eight cancers while MRI identified two – this difference, however, was not significant, the team noted.

Sore muscles, achy joints, and swollen toes – it’s not what we typically think of when we talk about COVID-19. But, these symptoms are proof that the virus does not only attack the body – it pushes the body to attack itself. In a study released this week in 

investigators revealed that CT, MRI and ultrasound scans show evidence of how the virus can kick-start rheumatological issues in some patients. By examining scans from patients who presented between May 2020 and December 2020, the team pinpointed several causes of these musculoskeletal symptoms, including edema, inflammation, hematomas, and gangrene. In addition, some images revealed blood clots or nerves that were bright and enlarged, indicating injury. If radiologists know to look for these findings, they can better identify patients who might benefit from a referral to a rheumatologist or a dermatologist for more extensive treatment, the team said.

 spoke with several of the radiologists from Penn Medicine, the Children’s Hospital of Pennsylvania, the University of Texas Southwestern, and the University of California at San Francisco who are currently involved in efforts to expand radiology education and technologies to resource-limited areas globally. They discussed the importance of these efforts, what has been accomplished to date, and what the future holds. Here’s what they shared.

Videos

Mammography "Sweet Spot" Recall Rate; MRI for Early-Stage Testicular Cancer Follow-Up; COVID-19 and Body Self-Attack; Plus, Global Radiology

Diagnostic Imaging's Weekly Scan: Feb. 15, 2021, to Feb. 19, 2021

A new X-ray imaging scanner could make it easier for breast surgeons to determine if they have removed all cancerous tissue during a breast-conserving surgery.

While conserving surgeries – those that remove only the tumor from breast cancer patients rather than removing the entire breast – account for most breast cancer operations, they are not always completely effective. In many instances, a second surgery is needed because it is difficult for surgeons to know for certain if they have removed all malignant tissue at the time of surgery.

However, a group of researchers from University College London, Queen Mary University of London, Barts Health NHS Trust, and Nikon have developed a new new X-ray imaging approach that lets surgeons determine – during initial surgery – if they have cleared the breast of cancer. This method, they said, improves detection of diseased tissue at the margins 2.5-times over standard imaging.

Examples of the imaging performance of XPCI-CT (b,e) compared to conventional specimen radiography (a,d) and benchmarked against histopathology (c,f). he top row focuses on the similarity between the XPCI-CT slice in (b) and the histological slice in (c). Arrow 1 indicates margin involvement, arrow 2 a variation in density in the internal structure of the tumor mass, arrow 3 tumor-induced inflammation. All this is confirmed by the histological slice in (c), and hardly visible in the conventional image in (a). The bottom row focuses on the detection of small calcifications, a key feature in DCIS. These are undetectable in (d), detected in (e), enhanced in the maximum intensity projection (MIP) image at the bottom of (f), and confirmed by histopathology in the top part of (f). The scale bar [shown in (b) and (e)] is the same for all images apart from (f), which has its own scale. Red arrows in (e) and (f) indicate the microcalcifications.

They published their findings on Feb. 11 in 

In the paper, they discussed how using X-ray phase contrast imaging (XPCI) for phase-based CT can give surgeons a full 3D image of an extracted tissue lump, also known as a wide local excision (WLE). Beyond assisting the surgeon, the researchers said, if implemented, the technology also offers benefits to the patient.

“This technology has the potential to significantly improve radiological intraoperative assessment of tumor margins,” said co-author Louise Jones, Ph.D., professor and director of the Breast Cancer Now Tissue Bank, “potentially reducing the need for repeat operations that many patients require, which can cause significant distress to the patient.”

X-ray Signal Extraction Method Could Be Next Generation Breast Cancer Screening Technique

According to the team, XPCI offers better soft tissue sensitivity than a conventional X-ray. Rather than detecting a change in an X-ray beams intensity as it moves through tissue, XPCI measures speed changes. In that way, it augments the contrast of soft tissues, including breast tumors.

Currently, WLE histopathological results are only available after a few days, so surgeons – and patients – must wait to find out if a procedure was successful in removing all cancerous tissue. If any affected margins remain, patients must undergo a second surgery.

Knowing whether the entire tumor has been excised before concluding the initial surgery is preferable, the team said, but, to-date, all attempts at intraoperative assessment have failed. Either these strategies have been unable to detect all diseased tissue, they could not penetrate deeply enough into the specimen to provide an accurate assessment, or they could do neither.

To determine the capabilities and efficacy of XPCI, the team tested the scanner with 101 WLEs and compared the outcomes to those from standard imaging, based on conventional X-rays, that was used intraoperatively. Half of the examined specimens contained a neoplastic lesions. Based on their evaluation, the team found XPCI had much higher sensitivity than conventional imaging – 83 percent and 32 percent, respectively. Specificity rates were comparable – 83 percent to 86 percent, respectively.

Overall, they said, the 2.5-times improvement they saw in detection could also translate into a similar drop in the re-operation rate.

“Although we have focused on the impact of XPCI in breast surgery,” said Tamara Suaris, consultant breast radiologist at Barts NHS Health Trust, “there is a much wider clinical potential of this technology, including other intraoperative areas, such as intestinal, esophageal, and prostatic surgery, and in the longer term, diagnostic imaging, notably mammography.”

This development opens the door for the first potential clinical use for XPCI, said lead author Alessandro Olivo, Ph.D., professor from UCL medical physics and biomedical engineering.

“The technology has tremendous potential,” he said, “and I am sure once people see what it can do many other clinical areas will follow suit.”

New strategy can better detect any remaining malignant tissue intraoperatively, increasing surgeon confidence and decrease patient anxiety.

New X-ray Scanner Could Reduce Repeat Breast-Conserving Operations

Case submitted by Saeed Rad, M.D., radiologist, to

 Patient presents with a history of excessive alcohol (ETOH) intake with increasing difficulty swallowing (dysphagia).

Esophageal varices are seen in over half of patients with portal hypertension, such as in the case of long standing alcohol intake. This is such a case. It is important to note there are other causes of cirrhosis and portal hypertension beyond alcohol intake, including Chid-Pugh C cirrhosis or other causes of liver disease that result in cirrhosis. 

Esophageal varices are dilated submucosal veins of the esophagus and are portosystemic collaterals. Fluoroscopic evaluation with barium demonstrates longitudinal and curvilinear luminal filling defects, representing varices.

What is your diagnosis in this patient with long standing excessive alcohol intake and difficulty swallowing?

Image IQ Quiz: Patient Presents with Increasing Difficulty Swallowing

Roughly 50 percent of patients who recover after being hospitalized with severe COVID-19 suffer heart damage. MRI scans, conducted at least one month post-discharge, show patients who have elevated levels of troponin show evidence of heart injury.

The amount of troponin, a protein that is released into the bloodstream when the heart muscle is injured, typically rises when arteries are blocked or when the heart is inflamed. The damage seen in COVID-19 patients includes myocarditis, muscle scarring or infarction, ischemia, or a combination of all three, said researchers in a study published Fed. 17 in 

The findings pointing to heart damage from this study contribute to results in recent months about the presence of myocarditis in athletes, adding to the understanding of how COVID-19 affects heart function.

Cardiac MRI Shows Less Myocarditis in Athletes Post-COVID-19 Than Previously Reported

“Raised troponin levels are associated with worse outcomes in COVID-19 patients,” said lead researcher Marianna Fontana, cardiology professor at University College London. “Patients with severe COVID-19 disease often have pre-existing heart-related health problems, including diabetes, raised blood pressure, and obesity. During severe COVID-19 infection, however, the heart may also be directly affected.”

MRI scan of damaged heart. Blue means reduced blood flow, orange is good blood flow. In this figure the inferior part of the heart shows dark blue, so the myocardial blood flow is very reduced and the black and white angiography, which looks directly at the blood vessels, shows that the vessel which supplies the blood to this part of the heart is occluded. The 3 colored images are 3 different slices of the heart: the basal the mid and the apical slice.

Knowing exactly how the heart is damaged by the virus is difficult. With heart MRI scans, though, providers can see different injury patterns that allow for more accurate diagnoses and the development of more targeted treatments, she continued.

Based on clinical observations, patients who are hospitalized with severe COVID-19 do have elevated troponin levels during the critical phase of their illness, indicating that the body has initiated an intense immune response. To learn more, Fontana’s team examined 148 COVID-19 patients who were discharged up until June 2020 from six acute hospitals throughout London. All included patients had elevated troponin levels and underwent MRI scans after being discharged. The team compared patient results to those of a control group who had not contracted COVID-19, as well as a group of 40 healthy volunteers.

“We found evidence of high rates of heart muscle injury that could be seen on the scans a month or two after discharge,” she said. “Whilst some of this may have been pre-existing, MRI scanning shows that some were new, and likely caused by COVID-19.”

According to the team’s evaluation, functionality of the left ventricle – the chamber that pumps oxygenated blood to every part of the body – was normal in 89 percent of COVID-19-affected patients, but 80 patients (54 percent) did have scarring or heart muscle injury. Inflammation was responsible for tissue scarring or injury patterns in 39 patients (26 percent), and, ischemic heart disease, including infarction or ischemia, was responsible in 32 patients (22 percent). Both causes were present in nine patients (6 percent), and 12 patients (8 percent) had ongoing heart inflammation.

“Importantly, the pattern of damage to the heart was variable, suggesting that the heart is at risk of different types of injury,” she explained. “While we detected only a small amount of ongoing injury, we saw injury to the heart that was present even when the heart’s pumping function was not impaired and might not have been picked up by other techniques.”

There are concerns that injury in patients who had the most severe cases can increase the risk for heart failure in the future, she noted. But, more investigation is required.

Ultimately, the team said, these findings present two opportunities for improving patient care and outcomes. Knowing more about the patterns of injury can help providers find ways to side-step injury before it even occurs, such as implementing treatments for blood clots that might be present. In addition, being able to pick up on the effects of these injuries during convalescence could help providers identify which patients might benefit from specific heart-protective drugs.

It is important to note, Fontana said, while this work indicates potential methods for identifying low- and high-risk patients and points to potential ways to improve outcomes, their investigation focused only on patients with elevated troponin levels who had been hospitalized.

In addition to the journal publication, Fontana and colleagues will present their results at the Society for Cardiovascular Magnetic Resonance annual meeting on Feb. 19.

MRI scans reveal myocarditis, scarring, infarction, and ischemia in recovered patients with elevated troponin levels. 

MRI Shows Heart Damage in Nearly Half of Patients with Severe COVID-19

Radiologists can classify breast lesions more accurately if they use artificial intelligence algorithms fueled by 4D data captured with dynamic contrasted-enhanced MRI (DCE-MRI), new research has found.

investigators from the University of Chicago outlined their method for fusing the 4D volumetric and temporal details gleaned from DCE-MRI with 2D deep-learning breast lesions analysis. Using maximum intensity projection (MIP) of lesion features that are pulled from four dynamic time points during a DCE-MRI via a deep-learning algorithm can provide better lesion classification, they said.

“Incorporating 4D information in DCE-MRI by feature MIP in deep transfer learning demonstrated superior classification performance compared with using MIP images as input in the task of distinguishing between benign and malignant breast lesions,” said the team led by doctoral student Qiyuan Hu, noting that their method outperforms the existing MIP strategy that only uses one post-contrast subtraction image.

This new method is the second attempt by University of Chicago researchers to use 4D information from DCE-MRI. The first iteration, dubbed image MIP, collapsed suspicious 4D DCE-MRI areas of interest from second post-contrast subtraction images down into 2D MIP, and, its results were better than those rendered by 3D-based methods.

“Instead of collapsing the volumetric information at the image level to form MIP images, we do so at the feature level by taking the maximum of CNN features along the axial dimension for a given lesion directly within the deep neural network architecture,” they said.

To gather more substantive temporal information, this method – called feature MIP – incorporates subtraction images from four dynamic points in the DCE sequence that are, then, put into red, green, and blue CNN channels. The team postulated that this new method might perform better in distinguishing between benign and malignant breast lesions because it might better leverage DCE-MRI’s volumetric information.

They set out to prove their hypothesis by using a retrospective dataset of 1,990 distinct lesions – 1,494 malignant and 496 benign – from 1,979 women. With a training and validation set of 1,455 lesions gathered between 2015 and 2016, Hu’s team trained linear support vector machine classifiers to characterize lesions as benign or malignant based on features extracted from the CNN.

Based on their analysis, the team determined that, on an independent 535-lesion test set, image MIP and feature MIP produce areas under the curve of 0.91 and 0.93, respectively, with the feature MIP method outperforming image MIP.

The team said they plan to expand this method to T2-weighted and diffusion-weighted MRI sequences in multi-parametric MRI, using images gathered from other medical centers and groups. Doing so, they concluded, will allow them to evaluate how their method performs across manufacturers, protocols, and patient groups.

New method can help providers distinguish between benign and malignant breast lesions.

4D-Fueled AI with DCE-MRI Improves Breast Lesion Characterization

Before we get to our featured interview with Linda Blount, MPH, president and chief executive officer for Black Women’s Health Imperative about the inequities of access to digital breast tomosynthesis, here are the top stories of the week.

Adding a radiomics-based machine learning algorithm to CT colonography allows for the differentiation between benign and pre-cancerous polyps picked up on colorectal cancer screening. To date, radiologists have not been able to make that determination using CT colonography, but thanks to work from German 

, it is possible to do so with high sensitivity and specificity. They built a machine algorithm that uses radiomics to extract quantitative image features that predict whether a polyp will develop into cancer. They trained the algorithm on 169 CT segmentations from 107 colorectal polyps from 63 patients and tested it on 118 segmentations from 77 polyps from 59 patients. Using random forest analysis, they determined their algorithm could accurately differentiate benign and pre-malignant polyps with 82-percent sensitivity and 85-percent specificity with an area under the curve of 0.91. Ultimately, the team said, incorporating a machine learning-derived algorithm into CT colonography can make it an even better colorectal cancer screening tool.

New brain MRI images have shown that COVID-19 can create leukoencephalopathy, a rare condition that affects white brain matter. Currently, the industry does not have defining criteria for COVID-19 disseminated leukoencephalopathy (CRDL), so researchers from the University of Pennsylvania investigated whether identifying signs was possible, and they published their findings in the 

. After examining 2,820 COVID-19-positive patients who were hospitalized between March 2020 and June 2020, they determined 59 patients underwent brain MRI. Those scans revealed six patients (10.2 percent) had findings that suggested CRDL, such as extensive confluent or multifocal white matter lesions, micro-hemorrhages, diffusion restriction, and enhancement. Other patients also had evidence of multiple sclerosis and small vessel ischemic disease white matter lesions; acute, subacute, and chronic infarcts; abnormal basal ganglia signal from hypoxemia; and micro-hemorrhage associated with acute, subacute, and chronic infarcts. The team did note that their findings could overlap with findings for other conditions, but they maintained that radiologists must remain alert to potential presentations of CRDL.

While CT is typically used in the angiography suite, new research published in 

 shows that using MRI instead can help providers determine whether to continue to thrombectomy, the place stents, or to administer anti-thrombotic medications with patients who have suffered acute ischemic stroke. In a retrospective study of patients who received endovascular intervention for acute ischemic stroke between July 2019 and May 2020, a team from the University of California at San Francisco determined that of 47 patients who underwent stroke intervention, 12 had X-ray angiography with MRI. Based on their analysis, including MRI directly impacted treatment decision for all 12, including whether to proceed with mechanical thrombectomy, to place intracranial stents, or whether to administer antithrombotic medication. But, while these advancements show promise, other experts said that streamlining workflow to reduce transfer time between the angiography unit and the MRI is critical because, with stroke, “time in brain.”

 spoke with Linda Blount, MPH, president and chief executive officer for Black Women’s Health Imperative about recent 

 that reveals black women do not have equal access to digital breast tomosynthesis despite better cancer detection and lower recall rates associated with the screening. She discussed what these inequities mean for this patient group and what can be done to alleviate the disparities. Here’s what she had to say.

CT Colonography & Tumor Differentiation; COVID-19 & Leukoencephalopathy; MRI, the Angiography Suite, and Acute Ischemic Stroke; Plus, DBT, African American Women, & Decreased Access

Diagnostic Imaging's Weekly Scan: Feb. 22, 2021, to Feb. 26, 2021

The statistics speak volumes. Approximately 30.3 million American adults have been diagnosed with heart disease, according to the Centers for Disease Control and Prevention (CDC). Moreover, cardiovascular disease is the No. 1 cause of death in the United States — responsible for taking the lives of 659,041 people in 2019 alone.

U.S. cardiologists have their work cut out for them. The specialty ranks among the busiest and most demanding, with a large volume of cases, as well as the need to respond to emergencies quickly. Experts say the challenges of the next decades will be many as the prevalence of various heart diseases will remain high due to a range of factors. These include family history of heart disease, poor diet, physical inactivity, obesity, and smoking, for example.

With such a pressing task, cardiologists are seeking the best possible medical technologies to help them properly diagnose and treat patients and improve overall outcomes. Medical imaging has come a long way. Today’s cardiologists have cutting edge diagnostic imaging tools at their disposal, and that’s an important first step.

But, imaging is only as good as its availability or accessibility. Workflow challenges are among the biggest hurdles for today’s cardiologists. The inability for a clinician to see the studies he or she needs at any given moment can literally mean the difference between life and death for a patient.

Fortunately, there is a solution to the myriad of workflow obstacles in the cardiac care space. A robust enterprise imaging technology can offer speedy access to images. But, it can do a whole lot more, too, like increase efficiencies, enhance collaboration, boost patient engagement, and help drive better outcomes.

Cardiology Challenges Demand an Enterprise Solution

A specialty that currently treats more than 92 million Americans conducts an enormous amount of testing and imaging. Providing anywhere, anytime access to growing cardiology datasets is a tall order, and organizations are under pressure to make data secure and readily available, both inside and outside the organization.

Disparate imaging and reporting systems in various departments is often a challenge, too. To enhance collaborative care, cardiologists need to share images with radiologists and other specialists, but very often their viewing tools are different, impeding this process.

Add to this the fact that we live in an era of mergers and acquisitions, where uniting hospitals often brings legacy systems that don't connect with one another—making the sharing of images a cumbersome process.

Finally, in the cardiac care space, reporting can be burdensome. The need to manually populate increasing volumes of clinical information into reports creates a bottleneck on cardiologists’ clinical workflows.

The upshot: The traditional Cardiology PACS is no longer a sustainable tool. To enable the best possible cardiac care and support the entire healthcare organization, Cardiology PACS solutions must evolve to function as a comprehensive enterprise imaging system.

, a next-generation PACS visualization platform that unites radiology, cardiology, and enterprise imaging through a single, zero-footprint PACS viewer.

This best-in-breed technology supports interoperability and access to the whole patient record—and that’s making a difference in patient care. For cardio teams, it enables on-demand, secure access to the holistic patient record which is having a positive impact on both workflow efficiencies and clinical outcomes.

Minnesota-based North Memorial Health is one of the top-rated hospitals in the nation for cardiovascular care. North Memorial’s Heart & Vascular group is led by a distinguished team of cardiologists and cardiothoracic surgeons, ensuring professional expertise of the highest standard.

When it came to medical imaging and reporting efficiencies, the cardiac group struggled. As a large, multi-site health system, North Memorial’s workflow issues and pain points were many.

For starters, various departments were using disparate solutions—impeding smooth and speedy workflow. For example, the health system’s cath lab, echo lab, vascular lab and nuclear medicine department all connected to different imaging and reporting systems. In addition, the organization’s legacy application lacked reporting customization capabilities.

North Memorial knew it was time for a more modern solution, including enterprise imaging. The team opted to install Fujifilm’s 

, a next-generation server-side rendering solution designed specifically to help streamline image review and reporting across cardiovascular modalities.

Synapse Cardiology PACS was designed to improve cardiology PACS workflow. It was built with cardiologists—and cardiac patients—in mind. It includes a secure image management and reporting platform that meets the needs of busy cardiologists’ caseloads, affording them proper and efficient reporting to continuously advance the quality of patient care.

This powerful technology simultaneously streamlines cardiologists’ workflow, image review, and advanced reporting, eliminating bottlenecks and allowing more efficient care to flourish.

Enhanced Efficiencies, Collaboration, Patient Engagement

Addressing workflow problems with enterprise imaging technology really does have a direct link to improving cardiac care. The North Memorial team is experiencing many benefits thanks to Synapse Cardiology PACS.

The consolidation of all diagnostic cardiology images and data means cardiologists work more efficiently by having all relevant information on one platform at the time of reporting.

Advanced reporting templates are improving workflow, too. North Memorial physicians actually helped design templates that maximize efficiencies while they're reading—freeing up valuable time that can now be spent focusing on patient care.

And the server-side technology is exceptionally speedy. That means physicians can have their procedure report into the North Memorial Health EHR almost instantaneously.

Perhaps one of the greatest benefits is that Synapse Cardiology PACS displays the complete patient record and provides access across cardiology and radiology PACS—a merge of cardiology and radiology imaging designed to improve clinical care. In fact, this dual ability is one of the things that makes the system so unique.

For example, cardiologists can evaluate a patient’s ECG in Synapse Cardiology alongside a chest X-ray in Synapse Radiology PACS—leading to quicker treatment decisions.

Clinicians can work anywhere, anytime and immediately see images they need to interpret. Instead of being limited to the cath lab, for example, they can access images and create reports from home.

Equally noteworthy, this enterprise solution enhances collaboration in cardiac care. With a single viewer, all providers have access to and view the same holistic patient picture. This makes it easier for physicians to work together, no matter where they are located.

For better cardiology services, providers must look beyond the existing cardiology PACS.