A SYSTEM FOR EVALUATING THE EFFECTIVENESS OF NEOADJUVANT

CHEMORADIOTHERAPY IN PATIENTS WITH COLORECTAL CANCER BASED

ON A TEXTURE ANALYSIS OF POST-THERAPEUTIC T2-WI

MAGNETIC RESONANCE IMAGING

 

Berezovskaya T.P.1, Dayneko Ya.A.1, Nevolskikh A.A.1, Ivanov S.A.1, Kaprin A.D.2

Purpose. To develop and test scoring system based on a texture analysis (TA) of the post-therapeutic T2-weighted magnetic resonance images of tumors in order to divide patients with locally advanced rectal cancer (RC) into two groups, i.e., those who were responsive (R) and non responsive (NR) groups to neoadjuvant chemoradiotherapy (NCRT).

Materials and methods. A retrospective study included 64 patients with RC who received combined treatment with nCRT in А. Tsyb Medical Radiological Research Centre – branch of the National Medical Research Radiological Centre of the Ministry of Health of the Russian Federation from 2012 to 2017: the patients were divided into a training sample (42 persons) and a control (test) sample (22 – persons). A texture analysis (TA) was performed on post-therapeutic T2-weighted magnetic resonance images using the GLCM method and the MaZda ver.4.6 software to determine 11 texture parameters. In the training sample, 5 TA parameters were determined with cut off points for the R and NR patients, based on which a scoring system for evaluating the effectiveness of nCRT was created. The system was tested on the control sample with the definition of diagnostic efficiency parameters.

Results. Based on the training sample, 5 parameters were selected: AngScMom (p=0.000002); InvDfMom (p=0.0310); Entropy (p=0.000003); SumEntrp (p=0.0011) and DifEntrp, (p=0.0480) which significantly differed in the R and NR groups, with calculated cut off points included in the scoring system for evaluating the effectiveness of nCRT. If the value of the TA parameter matched to the level for R, then 1 score, with a sum 3 scores or more, the patient was assigned R, if the sum was less than 3 – to NR.

The scoring system based on these parameters had a sensitivity/specificity/predictive value of a positive result/predictive value of a negative result for the training and control samples 95%/83%/82%/95% and 83%/90%/91%/82%, respectively. The area under the ROC curve was 0,94 for the training sample and 0,92 for the control sample.

Conclusions. The texture analysis of MRI images is a potentially significant imaging biomarker for identifying patients with RC, both responsive and non-responsive to nCRT, which makes it possible to individualize subsequent surgical treatment.

 

1 - А. Tsyb Medical Radiological Research Centre – branch of the National Medical Research Radiological Centre of the Ministry of Health of the Russian Federation. Obninsk, Russia.

2 - National Medical Research Radiological Centre of the Ministry of Health of the Russian Federation.

Moscow, Russia.

Keywords: magnetic resonance imaging; rectal cancer, texture analysis, response assessment, radiomics, restaging.

 


Corresponding author:  Dayneko Ya.A.; e-mail: Этот e-mail адрес защищен от спам-ботов, для его просмотра у Вас должен быть включен Javascript

 

For citation: Berezovskaya T.P., Dayneko Ya.A., Nevolskikh A.A., Ivanov S.A, Kaprin A.D. A system for evaluating the effectiveness of neoadjuvant chemoradiotherapy in patients with colorectal cancer based on a texture analysis of post-therapeutic T2-WI magnetic resonance imaging. REJR 2020; 10(3):92-101. DOI:10.21569/2222-7415-2020-10-3-92-101.

.

Received:       26.05.20 Accepted:     19.08.20