Automated Artificial Intelligence Model Trained on a Large Data Set Can Detect Pancreas Cancer on Diagnostic Computed Tomography Scans As Well As Visually Occult Preinvasive Cancer on Prediagnostic Computed Tomography Scans.

BACKGROUND & AIMS: The aims of our case-control study were (1) to develop an automated 3-dimensional (3D) Convolutional Neural Network (CNN) for detection of pancreatic ductal adenocarcinoma (PDA) on diagnostic computed tomography scans (CTs), (2) evaluate its generalizability on multi-institutional public data sets, (3) its utility as a potential screening tool using a simulated cohort with high pretest probability, and (4) its ability to detect visually occult preinvasive cancer on prediagnostic CTs. METHODS: A 3D-CNN classification system was trained using algorithmically generated bounding boxes and pancreatic masks on a curated data set of 696 portal phase diagnostic CTs with PDA and 1080 control images with a nonneoplastic pancreas. The model was evaluated on (1) an intramural hold-out test subset (409 CTs with PDA, 829 controls); (2) a simulated cohort with a case-control distribution that matched the risk of PDA in glycemically defined new-onset diabetes, and Enriching New-Onset Diabetes for Pancreatic Cancer score ≥3; (3) multi-institutional public data sets (194 CTs with PDA, 80 controls), and (4) a cohort of 100 prediagnostic CTs (i.e., CTs incidentally acquired 3-36 months before clinical diagnosis of PDA) without a focal mass, and 134 controls. RESULTS: Of the CTs in the intramural test subset, 798 (64%) were from other hospitals. The model correctly classified 360 CTs (88%) with PDA and 783 control CTs (94%), with a mean accuracy 0.92 (95% CI, 0.91-0.94), area under the receiver operating characteristic (AUROC) curve of 0.97 (95% CI, 0.96-0.98), sensitivity of 0.88 (95% CI, 0.85-0.91), and specificity of 0.95 (95% CI, 0.93-0.96). Activation areas on heat maps overlapped with the tumor in 350 of 360 CTs (97%). Performance was high across tumor stages (sensitivity of 0.80, 0.87, 0.95, and 1.0 on T1 through T4 stages, respectively), comparable for hypodense vs isodense tumors (sensitivity: 0.90 vs 0.82), different age, sex, CT slice thicknesses, and vendors (all P > .05), and generalizable on both the simulated cohort (accuracy, 0.95 [95% 0.94-0.95]; AUROC curve, 0.97 [95% CI, 0.94-0.99]) and public data sets (accuracy, 0.86 [95% CI, 0.82-0.90]; AUROC curve, 0.90 [95% CI, 0.86-0.95]). Despite being exclusively trained on diagnostic CTs with larger tumors, the model could detect occult PDA on prediagnostic CTs (accuracy, 0.84 [95% CI, 0.79-0.88]; AUROC curve, 0.91 [95% CI, 0.86-0.94]; sensitivity, 0.75 [95% CI, 0.67-0.84]; and specificity, 0.90 [95% CI, 0.85-0.95]) at a median 475 days (range, 93-1082 days) before clinical diagnosis. CONCLUSIONS: This automated artificial intelligence model trained on a large and diverse data set shows high accuracy and generalizable performance for detection of PDA on diagnostic CTs as well as for visually occult PDA on prediagnostic CTs. Prospective validation with blood-based biomarkers is warranted to assess the potential for early detection of sporadic PDA in high-risk individuals.

Fully Automated Deep Learning Model to Detect Clinically Significant Prostate Cancer at MRI.

Background Multiparametric MRI can help identify clinically significant prostate cancer (csPCa) (Gleason score ≥7) but is limited by reader experience and interobserver variability. In contrast, deep learning (DL) produces deterministic outputs. Purpose To develop a DL model to predict the presence of csPCa by using patient-level labels without information about tumor location and to compare its performance with that of radiologists. Materials and Methods Data from patients without known csPCa who underwent MRI from January 2017 to December 2019 at one of multiple sites of a single academic institution were retrospectively reviewed. A convolutional neural network was trained to predict csPCa from T2-weighted images, diffusion-weighted images, apparent diffusion coefficient maps, and T1-weighted contrast-enhanced images. The reference standard was pathologic diagnosis. Radiologist performance was evaluated as follows: Radiology reports were used for the internal test set, and four radiologists' PI-RADS ratings were used for the external (ProstateX) test set. The performance was compared using areas under the receiver operating characteristic curves (AUCs) and the DeLong test. Gradient-weighted class activation maps (Grad-CAMs) were used to show tumor localization. Results Among 5735 examinations in 5215 patients (mean age, 66 years ± 8 [SD]; all male), 1514 examinations (1454 patients) showed csPCa. In the internal test set (400 examinations), the AUC was 0.89 and 0.89 for the DL classifier and radiologists, respectively (P = .88). In the external test set (204 examinations), the AUC was 0.86 and 0.84 for the DL classifier and radiologists, respectively (P = .68). DL classifier plus radiologists had an AUC of 0.89 (P < .001). Grad-CAMs demonstrated activation over the csPCa lesion in 35 of 38 and 56 of 58 true-positive examinations in internal and external test sets, respectively. Conclusion The performance of a DL model was not different from that of radiologists in the detection of csPCa at MRI, and Grad-CAMs localized the tumor. © RSNA, 2024 Supplemental material is available for this article. See also the editorial by Johnson and Chandarana in this issue.

Bounding box-based 3D AI model for user-guided volumetric segmentation of pancreatic ductal adenocarcinoma on standard-of-care CTs.

OBJECTIVES: To develop a bounding-box-based 3D convolutional neural network (CNN) for user-guided volumetric pancreas ductal adenocarcinoma (PDA) segmentation. METHODS: Reference segmentations were obtained on CTs (2006-2020) of treatment-naïve PDA. Images were algorithmically cropped using a tumor-centered bounding box for training a 3D nnUNet-based-CNN. Three radiologists independently segmented tumors on test subset, which were combined with reference segmentations using STAPLE to derive composite segmentations. Generalizability was evaluated on Cancer Imaging Archive (TCIA) (n = 41) and Medical Segmentation Decathlon (MSD) (n = 152) datasets. RESULTS: Total 1151 patients [667 males; age:65.3 ± 10.2 years; T1:34, T2:477, T3:237, T4:403; mean (range) tumor diameter:4.34 (1.1-12.6)-cm] were randomly divided between training/validation (n = 921) and test subsets (n = 230; 75% from other institutions). Model had a high DSC (mean ± SD) against reference segmentations (0.84 ± 0.06), which was comparable to its DSC against composite segmentations (0.84 ± 0.11, p = 0.52). Model-predicted versus reference tumor volumes were comparable (mean ± SD) (29.1 ± 42.2-cc versus 27.1 ± 32.9-cc, p = 0.69, CCC = 0.93). Inter-reader variability was high (mean DSC 0.69 ± 0.16), especially for smaller and isodense tumors. Conversely, model's high performance was comparable between tumor stages, volumes and densities (p > 0.05). Model was resilient to different tumor locations, status of pancreatic/biliary ducts, pancreatic atrophy, CT vendors and slice thicknesses, as well as to the epicenter and dimensions of the bounding-box (p > 0.05). Performance was generalizable on MSD (DSC:0.82 ± 0.06) and TCIA datasets (DSC:0.84 ± 0.08). CONCLUSION: A computationally efficient bounding box-based AI model developed on a large and diverse dataset shows high accuracy, generalizability, and robustness to clinically encountered variations for user-guided volumetric PDA segmentation including for small and isodense tumors. CLINICAL RELEVANCE: AI-driven bounding box-based user-guided PDA segmentation offers a discovery tool for image-based multi-omics models for applications such as risk-stratification, treatment response assessment, and prognostication, which are urgently needed to customize treatment strategies to the unique biological profile of each patient's tumor.

Extraskeletal Ewing Sarcoma from Head to Toe: Multimodality Imaging Review.

Extraskeletal Ewing sarcoma (EES) is a rare subtype in the Ewing sarcoma family of tumors (ESFT), which also includes Ewing sarcoma of bone (ESB) and, more recently, primitive neuroectodermal tumors. Although these tumors often have different manifestations, they are grouped on the basis of common genetic translocation and diagnosis from specific molecular and immunohistochemical features. While the large majority of ESFT cases occur in children and in bones, approximately 25% originate outside the skeleton as EES. Importantly, in the adult population these extraskeletal tumors are more common than ESB. Imaging findings of EES tumors are generally nonspecific, with some variation based on location and the tissues involved. A large tumor with central necrosis that does not cross the midline is typical. Despite often nonspecific findings, imaging plays an important role in the evaluation and management of ESFT, with MRI frequently the preferred imaging modality for primary tumor assessment and local staging. Chest CT and fluorine 18 fluorodeoxyglucose PET/CT are most sensitive for detecting lung and other distant or nodal metastases. Management often involves chemotherapy with local surgical excision, when possible. A multidisciplinary treatment approach should be used given the propensity for large tumor size and local invasion, which can make resection difficult. Despite limited data, outcomes are similar to those of other ESFT cases, with 5-year survival exceeding 80%. However, with metastatic disease, the long-term prognosis is poor. ©RSNA, 2022.

Focal Nodular Hyperplasia and Focal Nodular Hyperplasia-like Lesions.

Focal nodular hyperplasia (FNH) is a benign lesion occurring in a background of normal liver. FNH is seen most commonly in young women and can often be accurately diagnosed at imaging, including CT, MRI, or contrast-enhanced US. In the normal liver, FNH frequently must be differentiated from hepatocellular adenoma, which although benign, is managed differently because of the risks of hemorrhage and malignant transformation. When lesions that are histologically identical to FNH occur in a background of abnormal liver, they are termed FNH-like lesions. These lesions can be a source of diagnostic confusion and must be differentiated from malignancies. Radiologists' familiarity with the imaging appearance of FNH-like lesions and knowledge of the conditions that predispose a patient to their formation are critical to minimizing the risks of unnecessary intervention for these lesions, which are rarely symptomatic and carry no risk for malignant transformation. FNH is thought to form secondary to an underlying vascular disturbance, a theory supported by the predilection for formation of FNH-like lesions in patients with a variety of hepatic vascular abnormalities. These include abnormalities of hepatic outflow such as Budd-Chiari syndrome, abnormalities of hepatic inflow such as congenital absence of the portal vein, and hepatic microvascular disturbances, such as those that occur after exposure to certain chemotherapeutic agents. Familiarity with the imaging appearances of these varied conditions and knowledge of their association with formation of FNH-like lesions allow radiologists to identify with confidence these benign lesions that require no intervention. Online supplemental material is available for this article. ©RSNA, 2022.

Pancreatic steatosis on computed tomography is an early imaging feature of pre-diagnostic pancreatic cancer: A preliminary study in overweight patients.

BACKGROUND: The prevalence of pancreatic ductal adenocarcinoma (PDAC) is on the rise, driven by factors such as aging and an increasing prevalence of obesity and diabetes mellitus. To improve the poor survival rate of PDAC, early detection is vital. Recently, pancreatic steatosis has gained novel interest as a risk factor for PDAC. This study aimed to investigate if pancreatic steatosis on computed tomography (CT) is an early imaging feature in patients with pre-diagnostic PDAC. METHODS: A retrospective case-control study was performed. Patients diagnosed with PDAC (2010-2016) were reviewed for abdominal non-contrast CT-imaging 1 month-3 years prior to their diagnosis. Cases were matched 1:4 with controls based on age, gender and imaging date. Unenhanced CT-images were evaluated for pancreatic steatosis (pancreas-to-spleen ratio in Hounsfield Units <0.70) by a blinded radiologist and results were compared between cases and controls. RESULTS: In total, 32 cases and 117 controls were included in the study with a comparable BMI (29.6 and 29.2 respectively, p = 0.723). Pancreatic steatosis was present in 71.9% of cases compared to 45.3% of controls (Odds ratio (OR) 3.09(1.32-7.24), p = 0.009). Adjusted for BMI and diabetes mellitus, pancreatic steatosis on CT remained a significant independent risk factor for PDAC (Adjusted OR 2.70(1.14-6.58), p = 0.037). CONCLUSION: Pancreatic steatosis measured on CT is independently associated with PDAC up to three years before the clinical diagnosis in overweight patients. If these data are confirmed, this novel imaging feature may be used to identify high-risk individuals and to stratify the risk of PDAC in individuals that already undergo PDAC screening.

Hepatic Mucinous Cystic Neoplasm Versus Simple Biliary Cyst: Assessment of Distinguishing Imaging Features Using CT and MRI.

OBJECTIVE. The purpose of our study was to identify the imaging features that differentiate a hepatic mucinous cystic neoplasm (MCN) from a simple biliary cyst. MATERIALS AND METHODS. Surgically resected hepatic MCNs and simple biliary cysts over a 20-year period (October 29, 1997-January 23, 2018) with preoperative CT, MRI, or both were retrospectively identified. Included cases underwent histopathologic confirmation of diagnosis based on the 2010 World Health Organization criteria and blinded imaging review. Various imaging features, including cyst shape and septal enhancement, were assessed for performance. For septate cysts, the relationship of the septation to the cyst wall-that is, arising from the wall without an indentation versus arising from an external macrolobulation-was recorded. Statistical analysis was performed for the imaging features with the chi-square test. RESULTS. The study group comprised 22 hepatic MCNs and 56 simple biliary cysts. A unilocular hepatic cystic lesion was highly predictive of a simple biliary cyst (positive predictive value = 95.2%). The imaging feature of septations arising only from macro-lobulations was 100% specific for a simple biliary cyst on CT (p = 0.001). The presence of septations arising from the cyst wall without indentation was 100% sensitive for hepatic MCN but was only 56.3% specific on CT. Septal enhancement reached 100% sensitivity for hepatic MCN on MRI (p = 0.018). CONCLUSION. The presence of septations, relationship of the septations to the cyst wall, and septal enhancement were sensitive imaging features in the detection of hepatic MCN. The imaging feature of septations arising only from macrolobulations in the cyst wall was specific for simple biliary cysts on CT and helped differentiate simple biliary cysts from hepatic MCNs.

Magnetic resonance imaging features in 283 patients with primary biliary cholangitis.

OBJECTIVES: To evaluate magnetic resonance imaging (MRI) features of the liver in primary biliary cholangitis (PBC). METHODS: We conducted a multicenter retrospective review on 283 patients with PBC who underwent an MRI between 2007 and 2018. Patients with overlap syndromes were excluded. MRI studies were independently reviewed by two abdominal radiologists for liver morphology, signal intensity, postcontrast enhancement, and decompensation. Liver and spleen volumes and normalized liver apparent diffusion coefficient (nlADC) were also calculated. MRI features were correlated with fibrosis stage among a subset of patients who had a liver biopsy within 6 months (n = 72). RESULTS: The study population was comprised of 283 patients (89% females) and a mean ± SD age of 59.4 ± 11.8 years. Lymphadenopathy (78.1%), periportal hyperintensity (36.7%), and periportal halo sign (27.6%) were the most common features. A positive correlation was found between fibrosis stage and spleen size (r = 0.457, p < 0.001), spleen volume (r = 0.557, p < 0.001) and portal vein diameter (r = 0.287, p = 0.013), and a negative correlation with nlADC (r = - 0.332, p = 0.011). Fibrosis stage also correlated with the presence of surface nodularity (p < 0.001), periportal halo sign (p = 0.04), collaterals (p = 0.033), and splenomegaly (p = 0.002). No significant differences in nlADC values were found in different fibrosis stages. Spleen size and volume were significantly higher in patients with ascites and collaterals (< 0.001). The periportal halo sign was present only in patients with significant fibrosis. None of the MRI features significantly correlated with inflammation grade. CONCLUSIONS: In PBC, presence of periportal halo sign correlates with significant fibrosis. Heterogeneous T2W intensity, heterogeneous postcontrast enhancement, collaterals, spleen size, and spleen volume correlate with fibrosis stage and may be useful for predicting advanced fibrosis. KEY POINTS: • The presence of periportal halo sign is indicative for significant fibrosis in primary biliary cholangitis. • Liver parenchymal heterogeneous T2 signal intensity, heterogeneous postcontrast enhancement, collaterals, spleen size, and spleen volume correlate with fibrosis stages in PBC and may be useful for predicting advanced fibrosis.

Multimodality Imaging of Abdominopelvic Tumors with Venous Invasion.

A broad range of abdominal and pelvic tumors can manifest with or develop intraluminal venous invasion. Imaging features at cross-sectional modalities and contrast-enhanced US that allow differentiation of tumor extension within veins from bland thrombus include the expansile nature of tumor thrombus and attenuation and enhancement similar to those of the primary tumor. Venous invasion is a distinctive feature of hepatocellular carcinoma and renal cell carcinoma with known prognostic and treatment implications; however, this finding remains an underrecognized characteristic of multiple other malignancies-including cholangiocarcinoma, adrenocortical carcinoma, pancreatic neuroendocrine tumor, and primary venous leiomyosarcoma-and can be a feature of benign tumors such as renal angiomyolipoma and uterine leiomyomatosis. Recognition of tumor venous invasion at imaging has clinical significance and management implications for a range of abdominal and pelvic tumors. For example, portal vein invasion is a strong negative prognostic indicator in patients with hepatocellular carcinoma. In patients with rectal cancer, diagnosis of extramural venous invasion helps predict local and distant recurrence and is associated with worse survival. The authors present venous invasion by vascular distribution and organ of primary tumor origin with review of typical imaging features. Common pitfalls and mimics of neoplastic thrombus, including artifacts and anatomic variants, are described to help differentiate these findings from tumor in vein. By accurately diagnosing tumor venous invasion, especially in tumors where its presence may not be a typical feature, radiologists can help referring clinicians develop the best treatment strategies for their patients. ©RSNA, 2020.

Cancer Detection Rate and Abnormal Interpretation Rate of Prostate MRI in Patients With Low-Grade Cancer.

PURPOSE: The aim of this study was to evaluate the utility of cancer detection rate (CDR) and abnormal interpretation rate (AIR) in prostate MRI for patients with low-grade prostate cancer (PCa). METHODS: This three-center retrospective study included patients who underwent prostate MRI from 2017 to 2021 with known low-grade PCa (Gleason score 6) without prior treatment. Patient-level highest Prostate Imaging Reporting & Data System (PI-RADS®) score and pathologic diagnosis within 1 year after MRI were used to evaluate the diagnostic performance of prostate MRI in detecting clinically significant PCa (csPCa; Gleason score ≥ 7). The metrics AIR, CDR, and CDR adjusted for pathologic confirmation rate were calculated. Radiologist-level AIR-CDR plots were shown. Simulation AIR-CDR lines were created to assess the effects of different diagnostic performances of prostate MRI and the prevalence of csPCa. RESULTS: A total of 3,207 examinations were interpreted by 33 radiologists. Overall AIR, CDR, and CDR adjusted for pathologic confirmation rate at PI-RADS 3 to 5 (PI-RADS 4 and 5) were 51.7% (36.5%), 22.1% (18.8%), and 30.7% (24.6%), respectively. Radiologist-level AIR and CDR at PI-RADS 3 to 5 (PI-RADS 4 and 5) were in the 36.8% to 75.6% (21.9%-57.5%) range and the 16.3%-28.7% (10.9%-26.5%) range, respectively. In the simulation, changing parameters of diagnostic performance or csPCa prevalence shifted the AIR-CDR line. CONCLUSIONS: The authors propose CDR and AIR as performance metrics in prostate MRI and report reference performance values in patients with known low-grade PCa. There was variability in radiologist-level AIR and CDR. Combined use of AIR and CDR could provide meaningful feedback for radiologists to improve their performance by showing relative performance to other radiologists.

Cancer Detection Rate and Abnormal Interpretation Rate of Prostate MRI Performed for Clinical Suspicion of Prostate Cancer.

PURPOSE: To report cancer detection rate (CDR) and abnormal interpretation rate (AIR) in prostate MRI performed for clinical suspicion of prostate cancer (PCa). MATERIALS AND METHODS: This retrospective single-institution, three-center study included patients who underwent MRI for clinical suspicion of PCa between 2017 and 2021. Patients with known PCa were excluded. Patient-level Prostate Imaging-Reporting and Data System (PI-RADS) score was extracted from the radiology report. AIR was defined as number of abnormal MRI (PI-RADS score 3-5) / total number of MRIs. CDR was defined as number of clinically significant PCa (csPCa: Gleason score ≥7) detected at abnormal MRI / total number of MRI. AIR, CDR, and CDR adjusted for pathology confirmation rate were calculated for each of three centers and pre-MRI biopsy status (biopsy-naive and previous negative biopsy). RESULTS: A total of 9,686 examinations (8,643 unique patients) were included. AIR, CDR, and CDR adjusted for pathology confirmation rate were 45.4%, 23.8%, and 27.6% for center I; 47.2%, 20.0%, and 22.8% for center II; and 42.3%, 27.2%, and 30.1% for center III, respectively. Pathology confirmation rate ranged from 81.6% to 88.0% across three centers. AIR and CDR for biopsy-naive patients were 45.5% to 52.6% and 24.2% to 33.5% across three centers, respectively, and those for previous negative biopsy were 27.2% to 39.8% and 11.7% to 14.2% across three centers, respectively. CONCLUSION: We reported CDR and AIR in prostate MRI for clinical suspicion of PCa. CDR needs to be adjusted for pathology confirmation rate and pre-MRI biopsy status for interfacility comparison.

Imaging cholangiocarcinoma: CT and MRI techniques.

Cross-sectional imaging plays a crucial role in the detection, diagnosis, staging, and resectability assessment of intra- and extrahepatic cholangiocarcinoma. Despite this vital function, there is a lack of standardized CT and MRI protocol recommendations for imaging cholangiocarcinoma, with substantial differences in image acquisition across institutions and vendor platforms. In this review, we present standardized strategies for the optimal imaging assessment of cholangiocarcinoma including contrast media considerations, patient preparation recommendations, optimal contrast timing, and representative CT and MRI protocols with individual sequence optimization recommendations. Our recommendations are supported by expert opinion from members of the Society of Abdominal Radiology's Disease-Focused Panel (DFP) on Cholangiocarcinoma, encompassing a broad array of institutions and practice patterns.

Diffuse diseases of the myocardium: MRI-pathologic review of nondilated cardiomyopathies.

OBJECTIVE: This article will present correlation of the key radiologic findings with gross and microscopic pathology for the characterization of diffuse myocardial diseases using advanced imaging techniques. Our goal is to provide a focused and in-depth review of the pathophysiology underlying each entity and to emphasize the structural basis for the corresponding imaging characteristics. This article is limited to those disorders characterized by ventricular wall thickening without chamber dilatation, including hypertrophic cardiomyopathy, hypertensive cardiomyopathy, and cardiac amyloidosis. CONCLUSION: For the characterization of diffuse myocardial diseases using advanced imaging techniques, it is essential to understand the underlying pathologic changes in the heart. With these techniques, such as cardiac MRI, the various cardiomyopathies can be differentiated accurately, which may potentially obviate invasive testing and endomyocardial biopsy.

Diffuse diseases of the myocardium: MRI-pathologic review of cardiomyopathies with dilatation.

OBJECTIVE: In this radiologic-pathologic review of the cardiomyopathies, we present the pertinent imaging findings of diffuse myocardial diseases that are associated with ventricular dilatation, including ischemic cardiomyopathy, nonischemic dilated cardiomyopathy, cardiac sarcoidosis, and iron overload cardiomyopathy. CONCLUSION: Correlation of the key radiologic findings with gross and microscopic pathologic features is presented, to provide the reader with a focused and in-depth review of the pathophysiology underlying each entity and the basis for the corresponding imaging characteristics.

A narrative review of imaging for pancreas adenocarcinoma: staging, surgical considerations, and surveillance.

Background and Objective: Pancreas adenocarcinoma is a disease with dire prognosis. Imaging is pivotal to the diagnosis, staging, reassessment, surgical planning, and surveillance of pancreas cancer. The purpose of this paper is to provide the reader an overview of current imaging practices for pancreas adenocarcinoma. Methods: A literature search of original papers and reviews through 2022 was performed using the PubMed database. The most current American College of Radiology Appropriateness Criteria and National Comprehensive Cancer Network guidelines on pancreas cancer imaging were also included. Key Content and Findings: Multidisciplinary team care at a high-volume institution is instrumental to optimal patient management and outcomes. It is therefore important for all team members to be aware of imaging modality options, strengths, and challenges. Additionally, a high-level understanding of imaging findings is useful clinically. This manuscript provides a current overview of imaging modalities used in the identification and assessment of pancreas adenocarcinoma, including ultrasound, computed tomography, magnetic resonance imaging, and positron emission tomography. Imaging findings, including the expected and unexpected, are reviewed to give the novice imager a better understanding. Conclusions: This review provides a current overview of imaging for pancreas adenocarcinoma, including strengths and weakness of various imaging modalities; therefore, providing the reader with a robust resource when considering imaging in the management of this disease.

Chemotherapy-associated liver morphological changes in hepatic metastases (CALMCHeM).

PURPOSE: To review imaging findings in chemotherapy-associated liver morphological changes in hepatic metastases (CALMCHeM) on computed tomography (CT)/magnetic resonance imaging (MRI) and its association with tumor burden. METHODS: We performed a retrospective chart review to identify patients with hepatic metastases who received chemotherapy and subsequent follow-up imaging where CT or MRI showed morphological changes in the liver. The morphological changes searched for were nodularity, capsular retraction, hypodense fibrotic bands, lobulated outline, atrophy or hypertrophy of segments or lobes, widened fissures, and one or more features of portal hypertension (splenomegaly/venous collaterals/ascites). The inclusion criteria were as follows: a) no known chronic liver disease; b) availability of CT or MRI images before chemotherapy that showed no morphological signs of chronic liver disease; c) at least one follow-up CT or MRI image demonstrating CALMCHeM after chemotherapy. Two radiologists in consensus graded the initial hepatic metastases tumor burden according to number (≤10 and >10), lobe distribution (single or both lobes), and liver parenchyma volume affected (<50%, or ≥50%). Imaging features after treatment were graded according to a pre-defined qualitative assessment scale of "normal," "mild," "moderate," or "severe." Descriptive statistics were performed with binary groups based on the number, lobar distribution, type, and volume of the liver affected. Chi-square and t-tests were used for comparative statistics. The Cox proportional hazard model was used to determine the association between severe CALMCHeM changes and age, sex, tumor burden, and primary carcinoma type. RESULTS: A total of 219 patients met the inclusion criteria. The most common primaries were from breast (58.4%), colorectal (14.2%), and neuroendocrine (11.0%) carcinomas. Hepatic metastases were discrete in 54.8% of cases, confluent in 38.8%, and diffuse in 6.4%. The number of metastases was >10 in 64.4% of patients. The volume of liver involved was <50% in 79.8% and ≥50% in 20.2% of cases. The severity of CALMCHeM at the first imaging follow-up was associated with a larger number of metastases (P = 0.002) and volume of the liver affected (P = 0.015). The severity of CALMCHeM had progressed to moderate to severe changes in 85.9% of patients, and 72.5% of patients had one or more features of portal hypertension at the last follow-up. The most common features at the final follow-up were nodularity (95.0%), capsular retraction (93.4%), atrophy (66.2%), and ascites (65.7%). The Cox proportional hazard model showed metastases affected ≥50% of the liver (P = 0.033), and the female gender (P = 0.004) was independently associated with severe CALMCHeM. CONCLUSION: CALMCHeM can be observed with a wide variety of malignancies, is progressive in severity, and the severity correlates with the initial metastatic liver disease burden.

Pancreatic Cystic Lesions: Next Generation of Radiologic Assessment.

Pancreatic cystic lesions are frequently identified on cross-sectional imaging. As many of these are presumed branch-duct intraductal papillary mucinous neoplasms, these lesions generate much anxiety for the patients and clinicians, often necessitating long-term follow-up imaging and even unnecessary surgical resections. However, the incidence of pancreatic cancer is overall low for patients with incidental pancreatic cystic lesions. Radiomics and deep learning are advanced tools of imaging analysis that have attracted much attention in addressing this unmet need, however, current publications on this topic show limited success and large-scale research is needed.

Clinical Implementation of an Artificial Intelligence Algorithm for Magnetic Resonance-Derived Measurement of Total Kidney Volume.

OBJECTIVE: To evaluate the performance of an internally developed and previously validated artificial intelligence (AI) algorithm for magnetic resonance (MR)-derived total kidney volume (TKV) in autosomal dominant polycystic kidney disease (ADPKD) when implemented in clinical practice. PATIENTS AND METHODS: The study included adult patients with ADPKD seen by a nephrologist at our institution between November 2019 and January 2021 and undergoing an MR imaging examination as part of standard clinical care. Thirty-three nephrologists ordered MR imaging, requesting AI-based TKV calculation for 170 cases in these 161 unique patients. We tracked implementation and performance of the algorithm over 1 year. A radiologist and a radiology technologist reviewed all cases (N=170) for quality and accuracy. Manual editing of algorithm output occurred at radiology or radiology technologist discretion. Performance was assessed by comparing AI-based and manually edited segmentations via measures of similarity and dissimilarity to ensure expected performance. We analyzed ADPKD severity class assignment of algorithm-derived vs manually edited TKV to assess impact. RESULTS: Clinical implementation was successful. Artificial intelligence algorithm-based segmentation showed high levels of agreement and was noninferior to interobserver variability and other methods for determining TKV. Of manually edited cases (n=84), the AI-algorithm TKV output showed a small mean volume difference of -3.3%. Agreement for disease class between AI-based and manually edited segmentation was high (five cases differed). CONCLUSION: Performance of an AI algorithm in real-life clinical practice can be preserved if there is careful development and validation and if the implementation environment closely matches the development conditions.