4382
Improvement of prostate cancer detection combining a computer aided diagnosis system to TRUS-MRI targeted biopsy.1Department of Radiology, Sapienza University, Policlinico Umberto I, Rome, Italy
Synopsis
To validate the role of mpMRI combined to CAD system, to increase prostate cancer detection rate using TRUS-MRI guided biopsy. 167 individuals, with elevated PSA level and no previous positive biopsy were enrolled and 63 underwent targeted biopsy. Two radiologists evaluated the exams adopting PIRADSv2 and CAD system. Radiologists’ evaluation proved better diagnostic performance compared to CAD. The highest detection rate for clinically significant cancer was obtained biopsying “target into target” lesions. CAD system proved to be useful in pinpointing the neoplastic area within MRI lesions, representing a valuable tool in identifying biopsy targets to improve CDR.
Introduction
Up to date, several studies have demonstrated how TRUS-MRI targeted biopsy has significantly increased cancer detection rate (CDR) compared to standard biopsy; 1–3 notably, increasing the sensitivity in identifying clinically significant PCa and decreasing overdiagnosis of clinically non-significant PCa. 4 Different studies published in literature have reported how CAD software, incorporating different MR imaging techniques, including T2WI, DWI and DCE, may help less-experienced radiologists to differentiate benign from malignant prostate gland disorders. 5,6 Considering the background, the purpose of our study was to validate the role of mpMRI combined to CAD system (WATSON Elementary), to increase the detection rate of prostate cancer (PCa) using TRUS-MRI fusion targeted biopsy as reference test for histopathology validation.Methods
A cohort of 167 patients, with elevated PSA level (> 2.5 ng/mL) and no previous positive biopsy, were enrolled in the study from October 2016 to June 2017 with waiver of informed consent. Patients underwent diagnostic 3T MpMRI (T2WI on axial and coronal planes, DWI with six b values up to 1500 s/mm2, DCE images). Targeted biopsy was performed in 63 patients. Two expert urogenital radiologists evaluated the exams adopting PIRADSv2 (Radiologist Evaluation-RAD) and Watson Elementary CAD system (CAD Evaluation-CAD). The latter is a workflow-based viewing tool that employs an algorithm based on a voxel per voxel correlation to develop a colorimetric malignancy map the so-called Malignancy Attention Index (MAI). MAI value ranges from 0 to 1, with 0 and 1 indicating the lowest (blue) and highest (red) probability of neoplastic transformation, respectively. The MRI data were compared to TRUS-MRI fusion targeted biopsy pathology reports and to histopathologic whole-mount slides, when radical prostatectomy was performed.Results
To evaluate the diagnostic performance of the CAD system, Receiver Operating Characteristics (ROC) curves were computed for both the radiologist evaluation and the CAD analysis, using MRI/US-fusion targeted biopsy as reference test for histopathology confirmation. We used Wilcoxon signed-rank test for the hypothesis test. We considered a P≤0.05 as statistically significant. Youden's J statistic was made to select the best MAI cut-off value. Radiologists’ evaluation (RAD) proved better diagnostic performance compared to CAD system ([AUC=0,71 (0,63-0,79)] vs [AUC=0,63 (0,54-0,71)]) with the cancer detection rate being 35.8% and 68.6%, respectively. The highest detection rate was 81.81% for clinically significant cancer, obtained when biopsying “target into target” lesions. Overall cancer detection rate was 60.37%.Discussion
In the study, we have compared the detection rate of PCa obtained by the CAD system alone, the radiologist evaluation (RAD) and the two combined, by means of TRUS-MRI fusion targeted biopsy pathology reports and to histopathologic whole-mount slides, when radical prostatectomy was performed. The CDRs were respectively: 36%, 61% and 81%. Statistical analysis demonstrated how CAD cannot be used as stand-alone tool, due to its lower performance. However, CAD may be helpful to a non-experienced radiologist to pay attention on specific parenchymal abnormalities that has to be analyzed by the human reader, to correctly assign a score of malignancy suspicion (PIRADS v2 score system). Considering that in some cases the CAD targets partially overlapped with the radiological ones, we defined a target sub-category the “target into target” area. Representing targets identified by both the radiologist (RAD) and the CAD system, that yield the highest CDR.Conclusion
In this study, the Radiologist’ Evaluation (RAD) was more efficient than Computer Aided Diagnosis Evaluation (CAD) in the diagnosis of prostate cancer. However, the CAD system proved to be useful in pinpointing the neoplastic area within MRI lesions. Accordingly, it may represent a valuable tool in identifying biopsy targets to improve CDR, specifically in terms of tumor per core percentage.Acknowledgements
No acknowledgement found.References
1. Siddiqui MM, Rais-Bahrami S, Turkbey B, George AK, Rothwax J, Shakir N, et al. Comparison of MR/Ultrasound Fusion–Guided Biopsy With Ultrasound-Guided Biopsy for the Diagnosis of Prostate Cancer. JAMA. 2015 Jan 27;313(4):390.
2. Radtke JP, Schwab C, Wolf MB, Freitag MT, Alt CD, Kesch C, et al. Multiparametric Magnetic Resonance Imaging (MRI) and MRI–Transrectal Ultrasound Fusion Biopsy for Index Tumor Detection: Correlation with Radical Prostatectomy Specimen. Eur Urol. 2016 Nov;70(5):846–53.
3. Schoots IG, Roobol MJ, Nieboer D, Bangma CH, Steyerberg EW, Hunink MGM. Magnetic Resonance Imaging–targeted Biopsy May Enhance the Diagnostic Accuracy of Significant Prostate Cancer Detection Compared to Standard Transrectal Ultrasound-guided Biopsy: A Systematic Review and Meta-analysis. Eur Urol. 2015 Sep;68(3):438–50.
4. Filson CP, Natarajan S, Margolis DJA, Huang J, Lieu P, Dorey FJ, et al. Prostate cancer detection with magnetic resonance-ultrasound fusion biopsy: The role of systematic and targeted biopsies: CaP Detection With MR-US Fusion Biopsy. Cancer. 2016 Mar 15;122(6):884–92.
5. Hambrock T, Vos PC, Hulsbergen–van de Kaa CA, Barentsz JO, Huisman HJ. Prostate Cancer: Computer-aided Diagnosis with Multiparametric 3-T MR Imaging—Effect on Observer Performance. Radiology. 2013 Feb;266(2):521–30.
6. Doi K. Computer-aided diagnosis in medical imaging: Historical review, current status and future potential. Comput Med Imaging Graph. 2007 Jun;31(4–5):198–211.
Figures
