Zhenyu Shu1, Songhua Fang1, Qin Ye2, Dewang Mao1, Hongfeng Cao3, Peipei Pang4,5, Xiangyang Gong6,7. 1. Department of Radiology, Zhejiang Provincial People's Hospital, Affiliated People's Hospital of Hangzhou Medical College, NO. 158 Shangtang Road, Hangzhou, Zhejiang, China. 2. Second Clinical College, Zhejiang Chinese Medical University, Hangzhou, China. 3. Department of Anorectal, Zhejiang Provincial People's Hospital, Affiliated People's Hospital of Hangzhou Medical College, Hangzhou, China. 4. Department of Radiotherapy, Zhejiang Provincial People's Hospital, Affiliated People's Hospital of Hangzhou Medical College, Hangzhou, China. 5. GE Healthcare China, Shanghai, China. 6. Department of Radiology, Zhejiang Provincial People's Hospital, Affiliated People's Hospital of Hangzhou Medical College, NO. 158 Shangtang Road, Hangzhou, Zhejiang, China. cjr.gxy@hotmail.com. 7. Institute of Artificial Intelligence and Remote Imaging, Hangzhou Medical College, Hangzhou, China. cjr.gxy@hotmail.com.
Abstract
PURPOSE: To explore the clinical feasibility of predicting the efficacy of neoadjuvant chemoradiotherapy (nCRT) for rectal cancer on the basis of texture analysis (TA) of T2-weighted imaging (T2WI). METHODS: The cohort for this prospective study comprised 136 patients with rectal cancer to be treated with nCRT, all of whom underwent three MR scans (pre-, early, and post-nCRT). Treatment efficacy was assessed on the basis of the outcomes of pathologic complete response (pCR) and non-pCR as determined by postoperative pathological examination. Extraction and analysis of texture features in T2WI of defined tumor regions were performed by AK software. Pre- and early-nCRT texture features were selected as potential predictors of outcomes by logistic regression analysis, and a prediction model for pCR was developed. A receiver operating characteristic (ROC) curve was used to assess the predictive power of texture features in pre- and early-nCRT images. RESULTS: Univariate logistic regression analysis demonstrated that the pre-nCRT features of energy, entropy, and skewness, and early-nCRT features of variance, kurtosis, energy, and entropy were independent predictors of pCR. A prediction model incorporating these predictors was constructed by multivariate logistic regression, The AUCs of pre-nCRT, early, and combined models were 0.751, 0.831, and 0.873, respectively; the sensitivities 66, 71, and 75%, respectively; and the specificities 87.22, 86.11, and 91.67%, respectively. CONCLUSIONS: TA of T2WI images can predict the efficacy of nCRT for rectal cancer, possibly providing a new marker of tumor biological response in clinical practice.
PURPOSE: To explore the clinical feasibility of predicting the efficacy of neoadjuvant chemoradiotherapy (nCRT) for rectal cancer on the basis of texture analysis (TA) of T2-weighted imaging (T2WI). METHODS: The cohort for this prospective study comprised 136 patients with rectal cancer to be treated with nCRT, all of whom underwent three MR scans (pre-, early, and post-nCRT). Treatment efficacy was assessed on the basis of the outcomes of pathologic complete response (pCR) and non-pCR as determined by postoperative pathological examination. Extraction and analysis of texture features in T2WI of defined tumor regions were performed by AK software. Pre- and early-nCRT texture features were selected as potential predictors of outcomes by logistic regression analysis, and a prediction model for pCR was developed. A receiver operating characteristic (ROC) curve was used to assess the predictive power of texture features in pre- and early-nCRT images. RESULTS: Univariate logistic regression analysis demonstrated that the pre-nCRT features of energy, entropy, and skewness, and early-nCRT features of variance, kurtosis, energy, and entropy were independent predictors of pCR. A prediction model incorporating these predictors was constructed by multivariate logistic regression, The AUCs of pre-nCRT, early, and combined models were 0.751, 0.831, and 0.873, respectively; the sensitivities 66, 71, and 75%, respectively; and the specificities 87.22, 86.11, and 91.67%, respectively. CONCLUSIONS: TA of T2WI images can predict the efficacy of nCRT for rectal cancer, possibly providing a new marker of tumor biological response in clinical practice.
Entities:
Keywords:
Magnetic resonance imaging; Neoadjuvant chemoradiotherapy; Rectal tumor; Texture analysis
Authors: Niels W Schurink; Lisa A Min; Maaike Berbee; Wouter van Elmpt; Joost J M van Griethuysen; Frans C H Bakers; Sander Roberti; Simon R van Kranen; Max J Lahaye; Monique Maas; Geerard L Beets; Regina G H Beets-Tan; Doenja M J Lambregts Journal: Eur Radiol Date: 2020-02-07 Impact factor: 5.315
Authors: Toru Tochigi; Sophia C Kamran; Anushri Parakh; Yoshifumi Noda; Balaji Ganeshan; Lawrence S Blaszkowsky; David P Ryan; Jill N Allen; David L Berger; Jennifer Y Wo; Theodore S Hong; Avinash Kambadakone Journal: Eur Radiol Date: 2021-10-13 Impact factor: 7.034
Authors: Charlems Alvarez-Jimenez; Jacob T Antunes; Nitya Talasila; Kaustav Bera; Justin T Brady; Jayakrishna Gollamudi; Eric Marderstein; Matthew F Kalady; Andrei Purysko; Joseph E Willis; Sharon Stein; Kenneth Friedman; Rajmohan Paspulati; Conor P Delaney; Eduardo Romero; Anant Madabhushi; Satish E Viswanath Journal: Cancers (Basel) Date: 2020-07-24 Impact factor: 6.639