Differentiation of benign from malignant solid renal lesions using CT-based radiomics and machine learning: comparison with radiologist interpretation.

PURPOSE: To assess the performance of a machine learning model trained with contrast-enhanced CT-based radiomics features in distinguishing benign from malignant solid renal masses and to compare model performance with three abdominal radiologists. METHODS: Patients who underwent intra-operative ultrasound during a partial nephrectomy were identified within our institutional database, and those who had pre-operative contrast-enhanced CT examinations were selected. The renal masses were segmented from the CT images and radiomics features were derived from the segmentations. The pathology of each mass was identified; masses were labeled as either benign [oncocytoma or angiomyolipoma (AML)] or malignant [clear cell, papillary, or chromophobe renal cell carcinoma (RCC)] depending on the pathology. The data were parsed into a 70/30 train/test split and a random forest machine learning model was developed to distinguish benign from malignant lesions. Three radiologists assessed the cohort of masses and labeled cases as benign or malignant. RESULTS: 148 masses were identified from the cohort, including 50 benign lesions (23 AMLs, 27 oncocytomas) and 98 malignant lesions (23 clear cell RCC, 44 papillary RCC, and 31 chromophobe RCCs). The machine learning algorithm yielded an overall accuracy of 0.82 for distinguishing benign from malignant lesions, with an area under the receiver operating curve of 0.80. In comparison, the three radiologists had significantly lower accuracies (p = 0.02) ranging from 0.67 to 0.75. CONCLUSION: A machine learning model trained with CT-based radiomics features can provide superior accuracy for distinguishing benign from malignant solid renal masses compared to abdominal radiologists.

Cystic artery velocity as a predictor of acute cholecystitis.

PURPOSE: To evaluate angle-corrected peak systolic cystic artery velocity (CAv) as a predictor of acute cholecystitis among patients presenting to the emergency department (ED) with right upper quadrant (RUQ) pain. METHODS: In this IRB-approved and retrospective study, CAv was evaluated in 73 patients, 43 who underwent definitive treatment with cholecystectomy or percutaneous cholecystostomy and 30 control patients without clinical suspicion for cholecystitis. In addition to CAv, the following were reviewed by 3 radiologists: CBD diameter, cholelithiasis, impacted stone in the neck, sludge, gallbladder wall thickness > 3 mm, gallbladder transverse dimension ≥ 4 cm, longitudinal dimension ≥ 8 cm, tensile gallbladder fundus sign, pericholecystic fluid, pericholecystic echogenic fat, and sonographic Murphy sign. RESULTS: Of the 43 patients who underwent definitive treatment, 25 had acute cholecystitis (34%) and 18 (25%) had chronic cholecystitis. Average CAv measurements were 50 ± 16 cm/s (acute), 28 ± 8 cm/s (chronic), and 22 ± 8 cm/s (control; p < 0.0001). In univariate analysis, among patients who underwent definitive therapy, CAv ≥ 40 cm/s, gallbladder wall thickness, stone impaction, GB long dimension ≥ 8 cm, and elevated WBC were associated with acute cholecystitis (p < 0.05). In multivariate analysis, CAv ≥ 40 cm/s was the only statistically significant variable (p = 0.016). CAv ≥ 40 cm/s alone had a PPV of 94.7% and overall accuracy of 81.4% in diagnosing acute cholecystitis. CONCLUSION: CAv ≥ 40 cm/s is highly associated with acute cholecystitis in patients presenting to the ED with RUQ pain.

Ileocolic vascular curvature: a new CT finding of cecal volvulus.

PURPOSE: To evaluate the diagnostic performance of a proposed new CT observation in cecal volvulus, marked hook-like curvature of ileocolic vessels termed "ileocolic vascular curvature." METHODS: Contrast-enhanced CT scans of 14 patients with diagnoses of cecal volvulus on original CT reports were compared with scans of 40 control patients with dilated cecum but no cecal volvulus on original CT reports, accrued consecutively from January, 2006 through July, 2017. Two independent blinded readers retrospectively evaluated scans for cecal dilatation, ileocolic vascular curvature and seven previously reported CT features of cecal volvulus. Statistical methods included the exact binomial distribution to define 95% confidence intervals (95% CIs), logistic regression, receiver operating characteristic analysis, and the exact McNemar test. Pathology reports and clinical records served as reference standards. RESULTS: All patients had abdominal pain and cecal distension. All 14 patients with CT diagnoses of cecal volvulus, and no control patients, had cecal volvulus by reference standards. Ileocolic vascular curvature and ectopic cecal location were the only features independently and significantly associated with cecal volvulus in multivariable regression (odds ratio 178, p = 0.014, and 63, p = 0.013, respectively) and also the only features with both sensitivity (12/14 [85.7%, 95% CI 57.2-98.2%] and 13/14 [92.9%, 95% CI 66.1-99.8%], respectively) and specificity (40/40 [100.0%, 95% CI 91.2-100.0%] and 38/40 [95.0%, 95% CI 83.1-99.4%], respectively) that differed significantly from 50.0%. CONCLUSIONS: Ileocolic vascular curvature was independently and significantly associated with cecal volvulus and exhibited both substantial sensitivity and substantial specificity for cecal volvulus, and thus is potentially a valuable new CT finding of cecal volvulus.

Groove pancreatitis: a clinical and imaging overview.

Groove or paraduodenal pancreatitis is an uncommon fibroinflammatory form of pancreatitis involving the anatomic space of the pancreatic groove located between the C-loop of the duodenum and the head of the pancreas. Although in some patients there are distinctive clinical and imaging features of groove pancreatitis (GP), there is often significant overlap with other infiltrative processes involving the pancreatic groove such as pancreatic ductal adenocarcinoma or duodenal carcinoma. In this review, we summarize the most distinctive clinical and imaging aspects of GP and highlight some important distinguishing features that may aid in differentiating malignant lesions involving the pancreatic groove.

Luminal obstruction in uncomplicated appendicitis: Detection with sonography and potential clinical implications.

PURPOSE: To determine the frequency of luminal obstruction on appendiceal graded compression sonography (US) in patients with uncomplicated appendicitis and test the hypothesis that this frequency is similar to the failure rates of primary antibiotic therapy for uncomplicated appendicitis when treatment selection is based on CT findings. METHODS: Sonograms of 150 consecutive patients with appendicitis diagnosed on ultrasound (US) and proven histopathologically were retrospectively evaluated; 114 (76.0%) had uncomplicated appendicitis, that is, appendicitis without gangrene or perforation, and were evaluated for appendicolithiasis and lymphoid hyperplasia using previously published sonographic criteria, and for luminal obstruction. RESULTS: Of the 114 patients with uncomplicated appendicitis, US demonstrated no luminal obstruction in 74 (64.9%) and luminal obstruction in 40 (35.1%, P = .018), the latter including 16 (40.0%) with lymphoid hyperplasia and 24 (60.0%, P = .074) with appendicolithiasis. CONCLUSIONS: US demonstrated luminal obstruction in 35.1% of patients with uncomplicated appendicitis, similar to the published failure rates of antibiotic therapy when treatment selection is based on CT. This confirms the hypothesis, supporting the possibility that undiagnosed luminal obstruction may account for treatment failures when patients are selected by CT for primary antibiotic therapy, and suggests a role for US in selecting patients without luminal obstruction for antibiotic therapy.