Role of T2 Textural Analysis of Prostate Lesion: A Retrospective Study

Abstract

Introduction: Prostate cancer is diagnosed in two-thirds of instances in the world's more developed regions. Prostate cancer was detected in 180,890 new cases in the United States in 2016, according to the American Cancer Society. One out of every six men is projected to develop prostate cancer at some point in their lives. The study's major purpose was to develop a textural analysis-based classifier to differentiate between benign and malignant prostate tumors using MRI-T2WI.

Materials and method: The retrospective study was conducted in the department of radiology in KKUH. Total 93 lesions from prostate cases were performed in KKUH from 2015 to 2017. About 75 lesions of 48 patients were included in this study. Eleven haralick features from region of intrests (ROIs) were extracted. After matching them with traces done by consultants in Profuse software, which was utilized for image-guided biopsy, digital rectal examination (DRE), prior biopsy (Prior bx) lesions were traced using ImageJ (MRI-ultrasound fusion). Weka software used this to create a classifier that distinguishes between malignant and benign tumors.

Result: The age of total 48 patients was in the interquartile range of 59.0-70.0, with an average of 64.4 years. The PSA was observed an average of 22.5 with an SD of 50.5 and an interquartile of 10.0. The mean size of the prostrate was 3.2 cm with SD 1.9. Among 48 patients Digital rectal examination (DRE) 8 (16.7%) and 40 (83.3%), Prior biopsy (PRIOR BX) 2 (4.2%) and 46 (95.8%), PI-RADS 22 (45.8%) and 26 (54.2%) were observed positive and negative respectively. In DRE, 88% sensitivity 55% specificity with PSA 9.75 (p-value 0.008) were observed. 100% of sensitivity, 41% specificity with PSA 8.19 (p-value 0.897) were found in PRIOR BX, but in MRI, 55% of sensitivity 69% specificity with PSA 10.70 (p-value 0.107) were observed.

Conclusion: T2 texture analysis is good in classifying prostate lesions with acceptable sensitivity and specificity. T2W MRI-based textural analysis agreed with pathological findings from many institutions and was sensitive to underlying pathological differences between low- and intermediate/high-grade prostate cancers. Actors in the diagnostic performance, such as DWI/ADC and perfusion, histogram parameters, and other features with distinct orientations and lengths, could help doctors discriminate benign and malignant prostate nodules, allowing for more efficient and precise clinical decisions.