Author: Logan Young

Since gender incongruence is now categorized as a sexual health condition, Florence X. Doo, Alexander S. Somwaru, and colleagues at Mount Sinai West in New York City contend that all subspecialties must be prepared to identify radiologic correlates and distinguish key postoperative variations in the three major categories of gender affirmation surgery: genital reconstruction, body contouring, and maxillofacial contouring. For trans-females, pelvic MRI remains the most reliable modality to evaluate the two most common complications arising from vaginoplasty: hematomas and fluid collection. Although vaginoplasty typically preserves the prostate, it may have atrophied from adjuvant hormonal therapy with estrogen and progesterone, so regular prostate cancer screening guidelines should still be followed. When evaluating urethral complications from phalloplasty in trans-males, for confirmation of stricture with abnormal function tests and for fistula evaluation, a retrograde urethrogram or voiding cystourethrogram can be obtained.

Should a patient desire erectile potential with the fully-healed neophallus, an implant may be placed, which is prone to infection, attrition, malposition, and constituent separation. Related to gender affirmation surgery, silicone or saline breast implants in trans-females often evidence as incidental notations on chest radiography, CT, and MRI, yet the most common body contouring gender affirmation surgery is subcutaneous mastectomy. Since the nipple-areola complex is preserved, retaining malignant transformation risk, Doo et al. recommend trans-males submit to regular postsurgical breast cancer screening. Likewise, trans-female patients who have undergone neoadjuvant hormone replacement therapy have an increased risk for breast cancer and should be routinely screened. Illegal silicone injections, long targeted toward all transgender populations, typically register incidentally on imaging studies, as do facial augmentations achieved via neurotoxin injections or fillers, such as calcium hydroxylapatite or hyaluronic acid. As Doo and Somwaru explain, “postoperative imaging is not typically obtained because external aesthetic results can be adequately evaluated by the surgeon,” unless unique complications—bony erosions from impaction of alloplastic silicone prostheses or bone and cartilage autografts, embolization from injection or filler materials, etc.—present themselves.

Collaborative research has not only standardized the definition of a complete imaging history, but also engineered systems to include supportive prompts in the order entry interface with a single keystroke—sustainably improving the overall quality of imaging histories. A Kaiser Permanente multidisciplinary team—physicians, both primary care and medical imaging providers, as well as information technology and practice improvement professionals—first defined the various components of a complete imaging history, a process that underwent several improvement cycles where consensus audits were regularly performed. The final apparatus of the collaborative team’s complete imaging history definition included the following responses: what happened; when it happened; where the patient was experiencing pain; and the ordering provider’s concern. These four prompts were then inserted into the electronic physician order entry process, and performance was monitored for an additional 18 months. From March 13, 2017, to December 16, 2018, 10,236 total orders were placed by ordering providers in the study clinic. Of the orders audited in the baseline period, 16.0% (64/397) contained all four history components, which increased to 52.0% (2200/4234; absolute increase of 36.0%, relative increase of 225.0%; p < 0.0001) in the subsequent time periods. Moreover, the mean number of characters ordering providers entered into the imaging histories they submitted increased from 45.4 characters per order during the baseline period to 75.4 (66.1% increase, p < 0.0001) after the intervention. “We have also noticed that increased clinical information results in radiologists feeling less of a need to recommend additional downstream or repeat imaging,” wrote Daniel S. Bor from Kaiser Permanente’s department of medical imaging. Bor noted, too, that the project team remains optimistic that the emergence of artificial intelligence could facilitate an automated method to improve imaging histories and support high-quality radiologist interpretations

With U.S. sales of false eyelashes having increased 31% since 2017 and magnetic eyelashes trending as the top beauty-related Google search of 2018, Einat Slonimsky and Alexander Mamourian at Penn State Health utilized a phantom to show that magnetic eyelashes worn during MRI can cause substantial artifact and that detachment of the eyelashes from the phantom can occur. Using two sets of magnetic eyelashes from the same manufacturer that were randomly selected and purchased online, the phantom was created by drilling multiple 2-mm holes in a plastic container and then running monofilament line through these holes to create a grid. The two sets of eyelashes were attached to single nylon strings, placed diagonally within the phantom. The phantom was then submerged in a container filled with distilled water, covered with a layer of plastic film to prevent free movement of the lashes, should they detach. MRI was performed using a 3-T scanner with T2-weighted images, FLAIR images, T1-weighted images, susceptibility-weighted images, DW images, T1-weighted magnetization-prepared rapid-acquisition gradient-echo images, and T2-weighted sampling perfection with application-optimized contrasts using different flip-angle evolutions. Ultimately, the magnetic eyelashes evidenced an artifact much larger than that created by the control aneurysm clips (two made of cobalt alloy, one made of titanium) using the same sequences—measuring 7 × 6 cm maximal on susceptibility-weighted images, obscuring the entire phantom. Although the eyelashes stayed attached to the strings during the scan, upon removal of the phantom from the bore, one set of eyelashes detached from its string. Restrained by the plastic covering, it became attracted to the other eyelashes still attached to the phantom. “We strongly recommend inserting a line about magnetic eyelashes on the MRI safety questionnaire and adding stops in the screening system to prevent the entry of anyone with these lashes, including staff, into the MRI scanner room,” Slonimsky and Mamourian wrote.

Aspirin therapy is associated with both improved liver function test results and survival after transarterial embolization (TAE) for hepatocellular carcinoma (HCC), according to a retrospective review of 304 patients led by F. Edward Boas at Memorial Sloan Kettering Cancer Center in New York City. Among the 42 patients taking aspirin at the time of initial TAE for HCC, bilirubin level evidenced lower 1 day (0.9 vs 1.3, p < 0.001), 1 month (0.9 vs 1.2, p = 0.048), and 1 year (0.8 vs 1.0, p = 0.021) post-embolization. “Although the differences in liver function test results in the groups taking and not taking aspirin were small,” wrote Boas, “standard biochemical liver function tests are insensitive to early cirrhotic changes.” Clarifying further, Boas noted, “small changes in biochemical liver function test results might underestimate the degree of liver injury after embolization.” Whereas aspirin use indicated no disparity in initial response rate (88% vs 90% complete response or partial response, p = 0.59), median time to progression (6.2 vs 5.2 months, p = 0.42), initial site of progression (p = 0.77), or fraction of patients dying with disease progression (88% vs 89%, p = 1.00), the median overall survival period after TAE for HCC measured longer for the cohort taking aspirin (57 vs 23 months, p = 0.008). Despite comparable liver function, American Joint Committee on Cancer stage, comorbidities, and other clinical characteristics before embolization in both groups, because his study was retrospective, Boas acknowledged that a confounding variable may account for the improved survival among patients taking aspirin.

Pathologically, Parkinson disease is characterized by a loss of dopaminergic neurons in the substantia nigra pars compacta (SNpc) area of the brain, resulting in presynaptic nigrostriatal dopamine dysfunction. Whereas prior research into dopaminergic neuroimaging has illustrated the independent utility of neuromelanin MRI and dopamine transporter SPECT for evaluating the severity of Parkinson disease, Hiroto Takahashi and colleagues from Japan’s Osaka University Graduate School of Medicine aimed to explore the use of these two imaging biomarkers to quantify severity during the progression of Parkinson disease. Twenty men and 20 women (mean age 68.35 years) who underwent neuromelanin MRI and dopamine transporter SPECT were included in the study. Parkinson disease severity was assessed with the Hoehn and Yahr (HY) scale (HY stage 1, 4 patients; stage 2, 18 patients; stage 3, 8 patients; stage 4, 6 patients; stage 5, 4 patients). The signal-to-noise ratio (SNR) in the SNpc on neuromelanin MR images and the striatal specific binding ratio (SBR) on dopamine transporter SPECT images were calculated based on the value of each background region. The Mann-Whitney U test was used to assess the significance of differences between the early-stage group (HY 1 and 2) and the advanced-stage group (HY 3–5) for each SNR and SBR. Overall, both SNR and SBR measured much greater in early-stage patients compared to the advanced stage group (p < 0.05). Additionally, the AUC for differentiating early and advanced stage groups was 0.73 for SNR and 0.89 for SBR. The coefficient of correlation was −0.47 for SNR versus HY stage and −0.67 for SBR versus HY stage. As Takahashi acknowledged, “The current study shows that it is possible to quantify the degeneration of dopaminergic nigrostriatal transporters in Parkinson disease using striatal SBR derived from dopamine transporter SPECT with good correlation with the HY stage.”

One topic that has the potential to drastically improve the value of MRI (but has not been researched as thoroughly as hardware application and software innovation) is the optimization of MRI workflow. The construction of a new outpatient imaging center at NYU Langone Health to house 1.5- and 3-T scanners provided Dr. Michael Recht and colleagues an opportunity to reengineer their MRI turnaround time (i.e., “the interval between completion of the last sequence acquisition for one patient and initiation of the first sequence acquisition for the next patient”). A process improvement team comprised of different stakeholders—radiologists, MRI technologists, IT administrators, front desk personnel, real estate and development staff—met biweekly for several months to define the current workflow and its limitations and to determine optimizations, such as the use of dockable tables, dedicated patient preparation rooms, two doors in each MRI room, positioning the scanner to provide the most direct path to the scanner, and coil storage in the preparation rooms with duplication of the most frequently used coils. Once construction and a two-month training period were completed, mean and median turnaround times were measured for each scanner at NYU Langone’s new facility, as well as for six scanners with differing magnets at two existing outpatient centers from March 1, 2018 to June 30, 2018. When patients were properly prepared and on their docking tables, the authors found the difference in mean turnaround time was 328 seconds, “which is greater than the goal of five minutes of time saved,” Recht wrote. For all patients, including those who were not ready when the prior patient’s examination was finished, the improvement in mean turnaround time was 265 seconds. Interestingly, the difference in median turnaround times for all patients clocked in at 340 seconds. “Five minutes might not seem like much time,” Aunt Minnie added, “but taking 300 fewer seconds to turn around an MRI suite to accommodate the next case can improve workflow, better serve patients, and add to a facility’s coffers.”

The use of prostate multiparametric MRI (mpMRI) and targeted biopsy increases detection of clinically significant cancers while decreasing the diagnosis of indolent disease. “One obstacle to broad application of prostate mpMRI is the lack of standardization and training necessary to interpret mpMR images,” wrote Matthew D. Greer, MD, department of radiation oncology at the University of Washington School of Medicine in Seattle. To evaluate agreement among radiologists across experience levels in the detection and assessment of prostate cancer at mpMRI using Prostate Imaging Reporting and Data System version 2 (PI-RADSv2), Greer’s team evaluated a total study population of 163 patients. Of these, 110 underwent prostatectomy after MRI, and 53 evidenced normal MRI findings and transrectal ultrasound–guided biopsy results. Of the nine radiologists representing six countries and eight institutions who volunteered for the study, three had a high level of experience in prostate mpMRI (> 2000 cases in the last two years), three had a moderate level (500–2000 cases in the last two years), and three had a low level of experience (< 500 cases in the last two years). The authors’ results found that sensitivity for index lesions was comparable among all radiologists (average, 92.2%; p = 0.360), but specificity proved experience-dependent. Highly experienced readers had 84.0% specificity versus 55.2% for all others (p < 0.001), suggesting that the decision to perform biopsy be set at a lower threshold for novice prostate mpMRI radiologists and in centers where prostate MRI is an uncommon examination. As Radiology Business noted, “radiologists were blinded to prostate-specific antigen level and previous biopsy results, and providing such information could have improved results.”

MRI remains the most sensitive tool for detecting breast cancer, but cost and acquisition time continue to be deterrents for women at average risk. For patients at University Hospitals Cleveland Medical Center who received a digital mammography (DM) or digital breast tomosynthesis (DBT) screening examination, patients with heterogeneously dense or extremely dense breast tissue received their density notification and mammogram results by mail, alongside a pamphlet explaining a new supplemental screening option: abbreviated protocol MRI. Choosing to implement the EA1141 study protocol—localizer sequences, axial T2-weighted STIR sequence, axial T1-weighted sequence with fat saturation before and after IV administration of gadolinium contrast—a team led by Holly Marshall, MD maintained a schedule of three 10-minute examinations in a single one-hour time slot (i.e., the time allotted for a complete MRI exam). After presenting the concept of “Fast MRI” to her hospital’s leadership board of surgeons, pathologists, breast clinicians, and medical and radiation oncologists, Marshall created online educational resources for referring physicians and patients. Marshall’s colleagues noted that it took 10 months to establish pricing through their institution’s finance department, as abbreviated MRI is a self-pay procedure that cannot be billed to insurance because no CPT code exists for it. Relying upon radiology self-pay tests such as unenhanced lung cancer CT screening and cardiac scoring as precedents for cost determination, Marshall concluded that her institution’s price is “substantially lower than that of the full breast MRI protocol” and “many patients with high-deductible insurance plans have a lower out-of-pocket expense for [abbreviated protocol] breast MRI than for the full protocol.”

Utilizing a Markov model developed via TreeAge Pro software, Paulo Henrique Lima, An Tang, and a team of researchers at the University of Montreal simulated seven strategies for surveilling and diagnosing hepatocellular carcinoma (HCC) in patients with cirrhosis: ultrasound (US) for surveillance, CT for diagnosis; US for surveillance, complete MRI for diagnosis; US for surveillance, CT for inadequate or positive surveillance; US for surveillance, complete MRI for inadequate or positive surveillance; surveillance and diagnosis with CT, followed by complete MRI for inadequate surveillance; surveillance and diagnosis with complete MRI, followed by CT for inadequate surveillance; and surveillance with abbreviated MRI, followed by CT for inadequate surveillance or complete MRI for positive surveillance. For both compliance scenarios evaluated—optimal and conservative—the most cost-effective strategy was based upon a willingness-to-pay threshold of $50,000 (Canadian) per quality adjusted life year. Base-case analysis revealed that the most cost effective strategy when compliance was optimal was surveillance and diagnosis with CT, followed by complete MRI for inadequate surveillance. When compliance was conservative, the best option was surveillance with abbreviated MRI, followed by CT for inadequate surveillance or complete MRI for positive surveillance. Although sensitivity analyses supported the base-case analysis in the optimal compliance scenario, several parameters altered cost effectiveness in the conservative scenario. “North American guidelines currently recommend [US] surveillance every 6 months in patients at risk for HCC,” as noted by Health Imaging; however, Lima et al. pointed out that compliance is “suboptimal” with an overall rate of 52%.

Using optical colonoscopy as their reference standard for concordance, a trio of researchers from the United States and Italy demonstrated high positive predictive value for colorectal cancer screening using CT colonography (CTC), resulting in greater specificity and risk stratification for appropriate patient triage compared with other noninvasive screening tests. In an AJR online exclusive, University of Wisconsin School of Medicine and Public Health radiologist, Perry J. Pickhardt, evaluated the positive predictive value for detecting colorectal lesions using only CTC compared to common stool-based assessments like the fecal occult blood test, fecal immunochemical test, and stool DNA. Of the 877 men and 773 women (median age 59.7 years) with 2688 total CTC-detected lesions 6 mm or larger, the overall positive predictive values were 88.8% by polyp and 90.8% by patient. In addition, a positive CTC examination pointed to the likelihood of abnormal tissue growth, as per-patient positive predictive values were 72.3% (1193/1650) for any neoplasia 6 mm or larger and 38.8% (641/1650) for advanced neoplasia. Perhaps more importantly, as Aunt Minnie reported, “CTC enabled radiologists to identify critical information about the kind of condition behind each positive finding.” For example, only 5.8% of positive CTC studies at the CTC Reporting and Data System (C-RADS) C2 category will have advanced neoplasia at optical colonoscopy; however, this increases to 67.1% and 79.4% for the C3 and C4 categories, respectively. “Similarly,” wrote Pickhardt, “although none of the 781 patients whose CTC studies were positive at the C-RADS C2 category (the lowest level of positive) had cancer, the cancer rate increases to over 50% at the C4 category (the most concerning positive level). By comparison, there is generally no such weighted information available to physicians and patients for a positive stool-based test.”

Researchers have validated a first-of-its-kind machine learning– based model to evaluate immunohistochemical (IHC) characteristics in patients with suspected thyroid nodules, achieving “excellent performance” for individualized noninvasive prediction of the presence of cytokeratin 19, galectin 3, and thyroperoxidase based upon CT images. “When IHC information is hidden on CT images,” principal investigator Jiabing Gu explained, “it may be possible to discern the relation between this information and radiomics by use of texture analysis.” To assess whether texture analysis could be utilized to predict IHC characteristics of suspected thyroid nodules, Gu and colleagues from China’s University of Jinan enrolled 103 patients (training cohort– to-validation cohort ratio, ≈ 3:1) with suspected thyroid nodules who had undergone thyroidectomy and IHC analysis from January 2013 to January 2016. All 103 patients—28 men, 75 women; median age, 58 years; range, 33–70 years—underwent CT before surgery, and 3D Slicer v 4.8.1 was used to analyze images of the surgical specimen. To facilitate test-retest methods, 20 patients were imaged in two sets of CT series within 10–15 minutes, using the same scanner (LightSpeed 16, Philips Healthcare) and protocols, without contrast administration. These images were used only to select reproducible and nonredundant features, not to establish or verify the radiomic model. The Kruskal-Wallis test (SPSS v 19, IBM) was employed to improve classification performance between texture feature and IHC characteristic. Gu et al. considered characteristics with p < 0.05 significant, and the feature-based model was trained via support vector machine methods, assessed with respect to accuracy, sensitivity, specificity, corresponding AUC, and independent validation. From 828 total features, 86 reproducible and nonredundant features were selected to build the model. The best performance of the cytokeratin 19 radiomic model yielded accuracy of 84.4% in the training cohort and 80.0% in the validation cohort. Meanwhile, the thyroperoxidase and galectin 3 predictive models evidenced accuracies of 81.4% and 82.5% in the training cohort and 84.2% and 85.0% in the validation cohort, respectively. Noting that cytokeratin 19 and galectin 3 levels are high in papillary carcinoma, Gu maintained that these models can help radiologists and oncologists to identify papillary thyroid cancers, “which is beneficial for diagnosing papillary thyroid cancers earlier and choosing treatment options in a timely manner.” Ultimately, asserted Gu, “this model may be used to identify benign and malignant thyroid nodules.”