Can MRI radiomics distinguish residual adenocarcinoma from acellular mucin in treated rectal cancer?

Background: We explored the use of MRI T2-weighted imaging (T2WI) radiomics to help distinguish acellular mucin from cellular mucin (residual tumor) to inform patient management. Methods: This retrospective, multi-institutional study included consecutive patients with rectal adenocarcinoma containing mucin on restaging MRI from March 2012–January 2020. Radiologists segmented 3-mm–thick T2WI. Data were split into training (n = 122), validation (n = 40), and test (n = 42) sets. Sensitivity, specificity, PPV, NPV, and AUC of the developed radiomic models were investigated on the test set. Histopathology was the reference standard. Two radiologists independently reviewed 42 patients using a unique five-point Likert scale for the probability of cellular mucin. DeLong-Delong or McNemar's test was used to compare the AUC and diagnostic performance, respectively, of the radiomics model to that of the human readers. Results: Of 204 patients (mean age, 56.6 ± 13.3 years; 129 men and 75 women), 39/204 (19 %) had acellular mucin whereas 165/204 (81 %) had cellular mucin. The radiomics model demonstrated a sensitivity of 90 % (27/30, 95 % CI: 74–97 %), specificity of 83 % (10/12, 95 % CI: 55–95 %), PPV of 93 % (27/29, 95 % CI: 78–98 %), NPV of 77 % (10/13, 95 % CI: 50–92 %), and AUC of 0.84 (95 % CI: 0.65–0.99), whereas the human readers showed inferior sensitivities of 47 % (14/30) (95 % CI: 28–66 %) and 50 % (15/30) (95 % CI: 31–69 %) and AUCs of 0.58 (95 % CI: 0.41–0.74) (DeLong's p = 0.06 [95 % CI: −0.55, 0.01]) and 0.73 (95 % CI: 0.57–0.88) (DeLong's p = 0.43 [95 % CI: −0.39, 0.16]). Conclusion: The developed MRI radiomics model predicts cellular mucin post neoadjuvant chemoradiation with higher sensitivity than human readers.