In vivo MRI based prostate cancer localization with random forests and auto-context model

Abstract

Prostate cancer is one of the major causes of cancer death for men. Magnetic resonance (MR) imaging is being increasingly used as an important modality to localize prostate cancer. Therefore, localizing prostate cancer in MRI with automated detection methods has become an active area of research. Many methods have been proposed for this task. However, most of previous methods focused on identifying cancer only in the peripheral zone (PZ), or classifying suspicious cancer ROIs into benign tissue and cancer tissue. Few works have been done on developing a fully automatic method for cancer localization in the entire prostate region, including central gland (CG) and transition zone (TZ). In this paper, we propose a novel learning-based multi-source integration framework to directly localize prostate cancer regions from in vivo MRI. We employ random forests to effectively integrate features from multi-source images together for cancer localization. Here, multi-source images include initially the multi-parametric MRIs (i.e., T2, DWI, and dADC) and later also the iteratively-estimated and refined tissue probability map of prostate cancer. Experimental results on 26 real patient data show that our method can accurately localize cancerous sections. The higher section-based evaluation (SBE), combined with the ROC analysis result of individual patients, shows that the proposed method is promising for in vivo MRI based prostate cancer localization, which can be used for guiding prostate biopsy, targeting the tumor in focal therapy planning, triage and follow-up of patients with active surveillance, as well as the decision making in treatment selection. The common ROC analysis with the AUC value of 0.832 and also the ROI-based ROC analysis with the AUC value of 0.883 both illustrate the effectiveness of our proposed method.

Fully automatic segmentation of paraspinal muscles from 3D torso CT images via multi-scale iterative random forest classifications

“…First, both methods are based on random forest classification with auto-context model [25]. Second, both studies confirm the effectiveness of incorporating context features for refined segmentation, despite the fact that the method introduced in [21] is applied to multi-parametric prostate MR images while the present method is evaluated on torso CT data. The differences between these two methods, however, are also apparent.…”

“…It is worth to compare the method introduced in [21] with the present method. First, both methods are based on random forest classification with auto-context model [25].…”

“…The differences between these two methods, however, are also apparent. More specifically, due to the purpose of the study reported in [21], which aims to localize prostate cancer from in vivo MR images, the resolution of their data is relatively low, leading to small data dimension along the out of plane direction. For example, the highest resolution of the multi-parametric MR images used in [21] is 0.3125 × 0.3125 × 3 mm 3 .…”

“…2). It was demonstrated in [21] that the context features could encode the spatial constraints into the classification, thus improving the quality of the estimated tissue probability maps.…”

“…Inspired by the auto-context model [21,25], we propose to employ features derived from multi-source information, including the original torso CT images and later also the iteratively estimated and refined probability maps of the paraspinal muscles.…”