To develop and validate a radiomics model for evaluating pathologic complete response (pCR) to neoadjuvant chemoradiotherapy in patients with locally advanced rectal cancer (LARC). We enrolled 222 patients (152 in the primary cohort and 70 in the validation cohort) with clinicopathologically confirmed LARC who received chemoradiotherapy before surgery. All patients underwent T2-weighted and diffusion-weighted imaging before and after chemoradiotherapy; 2,252 radiomic features were extracted from each patient before and after treatment imaging. The two-sample test and the least absolute shrinkage and selection operator regression were used for feature selection, whereupon a radiomics signature was built with support vector machines. Multivariable logistic regression analysis was then used to develop a radiomics model incorporating the radiomics signature and independent clinicopathologic risk factors. The performance of the radiomics model was assessed by its calibration, discrimination, and clinical usefulness with independent validation. The radiomics signature comprised 30 selected features and showed good discrimination performance in both the primary and validation cohorts. The individualized radiomics model, which incorporated the radiomics signature and tumor length, also showed good discrimination, with an area under the receiver operating characteristic curve of 0.9756 (95% confidence interval, 0.9185-0.9711) in the validation cohort, and good calibration. Decision curve analysis confirmed the clinical utility of the radiomics model. Using pre- and posttreatment MRI data, we developed a radiomics model with excellent performance for individualized, noninvasive prediction of pCR. This model may be used to identify LARC patients who can omit surgery after chemoradiotherapy. .

Purpose To study the relationship between MRI-defined extramural venous invasion (EMVI) prior to treatment and prognosis in patients with locally advanced rectal cancer treated with neoadjuvant chemotherapy-radiation therapy followed by surgery. Materials and Methods This retrospective study included 517 patients with locally advanced rectal cancer evaluated from August 2008 to December 2014. Baseline and posttherapy MRI and follow-up data were retrieved for all patients. After training by using 328 cases with pathologic evaluation of EMVI after therapy, radiologists evaluated baseline MRI for EMVI status in addition to tumor size and characteristics, nodal status, and invasion of the mesorectal fascia. Reader reproducibility was determined by using κ coefficient. Kaplan-Meier curves and adjusted Cox models were used to determine the relationship of baseline MRI parameters to overall survival, metastasis-free survival, and local relapse-free survival. Results Among 517 patients, 335 (64.8%) were men; the mean age was 55.6 years ± 11.5 (standard deviation). At baseline, radiologists identified 259 of 517 (50%) patients with EMVI by using MRI. In adjusted analysis, EMVI and mesorectal fascial invasion at baseline MRI were predictors of metastasis-free survival (hazard ratio, 0.3 and 0.6; P ˂ .01 and P ˂ .02, respectively) and overall survival (hazard ratio, 0.5 and 0.5; P = .01 and P = .02, respectively). EMVI was the only factor associated with local relapse-free survival (hazard ratio, 0.3; P ˂ .01). The κ coefficient for determination of EMVI was 0.80. Conclusion Extramural venous invasion (EMVI) can be reliably evaluated with MRI. The presence of EMVI was associated with greater risk of local and distant tumor recurrence and overall death in patients with locally advanced rectal cancer treated with neoadjuvant chemotherapy-radiation therapy. © RSNA, 2018 Online supplemental material is available for this article.

No abstract

Objective: To develop and validate a radiomics model of diffusion kurtosis imaging (DKI) and T2 weighted imaging for discriminating pancreatic neuroendocrine tumors (PNETs) from solid pseudopapillary tumors (SPTs). Materials and Methods: Sixty-six patients with histopathological confirmed PNETs (n = 31) and SPTs (n = 35) were enrolled in this study. ROIs of tumors were manually drawn on each slice at T2WI and DWI (b = 1,500 s/mm 2) from 3T MRI. Intraclass correlation coefficients were used to evaluate the interobserver agreement. Mean diffusivity (MD) and mean kurtosis (MK) were derived from DKI. The least absolute shrinkage and selection operator regression were used for feature selection. Results: MD and MK had a moderate diagnostic performance with the area under curve (AUC) of 0.71 and 0.65, respectively. A radiomics model, which incorporated sex and age of patients and radiomics signature of the tumor, showed excellent discrimination performance with AUC of 0.97 and 0.86 in the primary and validation cohort. Moreover, the new model had better diagnostic performance than that of MD (P = 0.023) and MK (P = 0.004), and showed excellent differentiation with a sensitivity of 95.00% and specificity of 91.67% in primary cohort, and the sensitivity of 90.91% and specificity of 81.82% in the validation cohort. The accuracy of radiomics analysis, radiologist 1, and radiologist 2 for diagnosing SPTs and PNETs were 92.42, 77.27, and 78.79%, respectively. The accuracy of radiomics analysis was significantly higher than that of subjective diagnosis (P < 0.05). Conclusions: Radiomics model could improve the diagnostic accuracy of SPTs and PNETs and contribute to determining an appropriate treatment strategy for pancreatic tumors.

No abstract