Desmoid-type fibromatosis: Tumour response assessment using magnetic resonance imaging signal and size criteria.

INTRODUCTION: This study aimed to establish a specified magnetic resonance imaging (MRI) signal and size criterion for assessing the response of desmoid-type fibromatosis (DF). METHODS: This retrospective study included 129 patients with DF who received non-surgical therapy. All patients underwent pretreatment and 6-month-interval follow-up MRI for >3 years (6 follow-up visits). The correlation between signal grade and size was determined. Signal grade and size among three response groups (partial response [PR], stable disease [SD], progression disease [PD]) were compared. The specified signal and size criterion was established, used to assess tumour response at each follow-up, and compared with the reference. The Response Evaluation Criteria in Solid Tumours (RECIST)1.1 criterion at the end of the 3rd year was considered the reference. RESULTS: MRI signals were moderately correlated with size changes (r = -0.56 and -0.41 for T2 grade and contrast-enhanced T1 grade, respectively). Changes in T2 grade and size in the three response groups were significantly different (all p < 0.01). The signal and size criterion accurately predicted 95% of PR patients at 2nd follow-up and 81.2% of PD patients at the 3rd follow-up, while only 13.1% of PR and 56.3% of PD patients were predicted by RECIST1.1. However, the accuracy of the signal & size criterion for predicting SD was lower than that of RECIST1.1. CONCLUSIONS: MRI signal is useful in assessing the response of DF. Signal & size criterion can identify patients with PR and PD earlier than RECIST1.1.

Predicting Rectal Cancer Response to Neoadjuvant Chemoradiotherapy Using Deep Learning of Diffusion Kurtosis MRI.

Background Preoperative response evaluation with neoadjuvant chemoradiotherapy remains a challenge in the setting of locally advanced rectal cancer. Recently, deep learning (DL) has been widely used in tumor diagnosis and treatment and has produced exciting results. Purpose To develop and validate a DL method to predict response of rectal cancer to neoadjuvant therapy based on diffusion kurtosis and T2-weighted MRI. Materials and Methods In this prospective study, participants with locally advanced rectal adenocarcinoma (≥cT3 or N+) proved at histopathology and baseline MRI who were scheduled to undergo preoperative chemoradiotherapy were enrolled from October 2015 to December 2017 and were chronologically divided into 308 training samples and 104 test samples. DL models were constructed primarily to predict pathologic complete response (pCR) and secondarily to assess tumor regression grade (TRG) (TRG0 and TRG1 vs TRG2 and TRG3) and T downstaging. Other analysis included comparisons of diffusion kurtosis MRI parameters and subjective evaluation by radiologists. Results A total of 383 participants (mean age, 57 years ± 10 [standard deviation]; 229 men) were evaluated (290 in the training cohort, 93 in the test cohort). The area under the receiver operating characteristic curve (AUC) was 0.99 for the pCR model in the test cohort, which was higher than the AUC for raters 1 and 2 (0.66 and 0.72, respectively; P < .001 for both). AUC for the DL model was 0.70 for TRG and 0.79 for T downstaging. AUC for pCR with the DL model was better than AUC for the best-performing diffusion kurtosis MRI parameters alone (diffusion coefficient in normal diffusion after correcting the non-Gaussian effect [Dapp value] before neoadjuvant therapy, AUC = 0.76). Subjective evaluation by radiologists yielded a higher error rate (1 - accuracy) (25 of 93 [26.9%] and 23 of 93 [24.8%] for raters 1 and 2, respectively) in predicting pCR than did evaluation with the DL model (two of 93 [2.2%]); the radiologists achieved a lower error rate (12 of 93 [12.9%] and 13 of 93 [14.0%] for raters 1 and 2, respectively) when assisted by the DL model. Conclusion A deep learning model based on diffusion kurtosis MRI showed good performance for predicting pathologic complete response and aided the radiologist in assessing response of locally advanced rectal cancer after neoadjuvant chemoradiotherapy. © RSNA, 2020 Online supplemental material is available for this article. See also the editorial by Koh in this issue.

MRI evaluation of extramural vascular invasion by inexperienced radiologists.

OBJECTIVE: We proposed to determine whether the performance of inexperienced radiologists in determining extramural vascular invasion (EMVI) in rectal cancer on MRI can be promoted by means of targeted training. METHODS: 230 rectal cancer patients who underwent pre-operative chemoradiotherapy were included. Pre-therapy and post-therapy MR images and pathology EMVI evaluation were available for cases. 230 cases were randomly divided into 150 training cases and 80 testing cases, including 40 testing case A and 40 testing case B. Four radiologists were included for MRI EMVI evaluation, who were divided into targeted training group and non-targeted training group. The two groups evaluated testing case A at baseline, 3 month and 6 month, evaluated testing case B at 6 month. The main outcome was agreement with expert-reference for pre-therapy and post-therapy evaluation, the other outcome was accuracy with pathology for post-therapy evaluation. RESULTS: After 6 months of training, targeted training group showed statistically higher agreement with expert-reference than non-targeted training group for both pre-therapy and post-therapy MRI EMVI evaluation of testing case A and testing case B, all p < 0.05. Targeted training group also showed significantly higher accuracy with pathology than non-targeted training group for post-therapy evaluation of testing case A and testing case B after 6 months of training, all p < 0.05. CONCLUSION: The diagnostic performance for MRI EMVI evaluation could be promoted by targeted training for inexperienced radiologist. ADVANCES IN KNOWLEDGE: This study provided the first evidence that after 6 month targeted training, inexperienced radiologists demonstrated improved diagnostic performance, with a 20% increase in agreement with expert-reference for both pre-therapy and post-therapy MRI EMVI evaluation and also a 20% increase in or accuracy with pathology for post-therapy evaluation, while inexperienced radiologists could not gain obvious improvement in MRI EMVI evaluation through the same period of regular clinical practice. It indicated that targeted training may be necessary for helping inexperienced radiologist to acquire adequate experience for the MRI EMVI evaluation of rectal cancer, especially for radiologist who works in a medical unit where MRI EMVI diagnosis is uncommon.

Prediction of pathological nodal stage of locally advanced rectal cancer by collective features of multiple lymph nodes in magnetic resonance images before and after neoadjuvant chemoradiotherapy.

OBJECTIVE: To predict pathological nodal stage of locally advanced rectal cancer by a radiomic method that uses collective features of multiple lymph nodes (LNs) in magnetic resonance images before and after neoadjuvant chemoradiotherapy (NCRT). METHODS: A total of 215 patients were included in this study and chronologically divided into the discovery cohort (n=143) and validation cohort (n=72). In total, 2,931 pre-NCRT LNs and 1,520 post-NCRT LNs were delineated from all visible rectal LNs in magnetic resonance images. Geometric, first-order and texture features were extracted from each LN before and after NCRT. Collective features are defined as the maximum, minimum, mean, median value and standard deviation of each feature from all delineated LNs of each participant. LN-model is constructed from collective LN features by logistic regression model with L1 regularization to predict pathological nodal stage (ypN0 or ypN+). Tumor-model is constructed from tumor features for comparison by using DeLong test. RESULTS: The LN-model selects 7 features from 412 LN features, and the tumor-model selects 7 features from 82 tumor features. The area under the receiver operating characteristic curve (AUC) of LN-model in the discovery cohort is 0.818 [95% confidence interval (95% CI): 0.745-0.878], significantly (Z=2.09, P=0.037) larger than 0.685 (95% CI: 0.602-0.760) of the tumor-model. The AUC of LN-model in validation cohort is 0.812 (95% CI: 0.703-0.895), significantly (Z=3.106, P=0.002) larger than 0.517 (95% CI: 0.396-0.636) of the tumor-model. CONCLUSIONS: The usage of collective features from all visible rectal LNs performs better than the usage of tumor features for the prediction of pathological nodal stage of locally advanced rectal cancer.

Dynamic enhanced CT: is there a difference between liver metastases of gastroenteropancreatic neuroendocrine tumor and adenocarcinoma.

This study proposed to evaluate the feasibility of dynamic enhanced CT in differentiation of liver metastases of gastroenteropancreatic well-differentiated neuroendocrine tumors (GEP NETs) from GEP adenocarcinomas based on their characteristic features. CT images of 23 well-differentiated (G1 or G2) GEP NETs and 23 GEP adenocarcinomas patients with liver metastases were retrospectively reviewed. The distribution type, shape, intra-tumoral neovascularity, enhancement on hepatic artery phase, dynamic enhancement pattern and lymphadenopathy were subjective analyzed. Meanwhile, the size, number, CT value of tumor and adjacent normal liver parenchyma were measured and the metastasis-to-liver ratios were calculated objectively. Compared with GEP adenocarcinomas, the liver metastases of GEP NETs more frequently demonstrated a hyper enhancement on hepatic artery phase, washout dynamic enhancement pattern, absence of lymphadenopathy and higher metastasis-to-liver ratios on both hepatic artery phase and portal venous phase (P=0.017, P<0.001, P =0.038, P <0.001 and P =0.008, respectively). Logistic regression analysis showed that the dynamic enhancement pattern (P=0.012), and the metastasis-to-liver ratios on hepatic artery phase (P=0.009) were independent CT predictors for liver metastases of GEP NETs. The sensitivity and specificity of combing the two predictors in differentiation of liver metastases of GEP adenocarcinomas from GEP NET were 82.6% (19 of 23) and 91.3% (21 of 23), respectively. CT features are helpful in differentiating liver metastases of well-differentiated GEP NETs from that of GEP adenocarcinomas.

Intravoxel incoherent motion diffusion weighted MRI of cervical cancer - Correlated with tumor differentiation and perfusion.

OBJECTIVES: To investigate the value of parameters derived from IVIM model in grading of uterine cervical cancer and the relationship between perfusion parameters derived from IVIM and that from DCE-MRI. METHODS: Parameters of DWI (ADC, D, f, D*) and semi-quantitative parameters of DCE-MRI (Slop, Maxslop, CER, Washout, AUC90) were assessed in 24 female with cervical cancers. Except for ROIs encompassed all of the area of tumors in axial plane (A_all), ROIs on tumor edge (A_peri) and tumor center (A_central) were drawn. All of the parameters were compared among three pathology grades. Perfusion parameters derived from IVIM were correlated with that from DCE-MRI. RESULTS: For G1, G2 and G3 tumors, on tumor edge ADC=(1.03±0.11), (1.05±0.10), (0.90±0.05)×10(-3)mm(2)/s, D=(0.80±0.11), (0.78±0.07), (0.69±0.06)×10(-3)mm(2)/s, and f=(0.19±0.03), (0.22±0.02), (0.24±0.03). The differences among groups were significant (P<0.05). On tumor center, ADC=(0.90±0.10), (0.85±0.03), (0.80±0.07)×10(-3)mm(2)/s with significant differences (P=0.027). The other parameter, D and f of tumor center, as well as D* of all tumor areas, were of no statistic significance. Most of the DCE-MRI parameters negatively correlated with tumor volume. Although the correlation between f and slop was statistic significant, R=0.277 meant a negligible correlation. f had week correlation with Maxslop, CER and AUC90 (R=0.361, 0.400 and 0.405; P<0.001). D* showed no statistic significant correlation with all of the DCE parameters. CONCLUSION: IVIM model could possibly be used to evaluate tumor differentiation and perfusion, providing an alternative for DCE-MRI.