Multi-stage osteolytic spinal bone lesion detection from CT data with internal sensitivity control

Abstract

Spinal bone lesion detection is a challenging and important task in cancer diagnosis and treatment monitoring. In this paper we present a method for fully-automatic osteolytic spinal bone lesion detection from 3D CT data. It is a multi-stage approach subsequently applying multiple discriminative models, i.e., multiple random forests, for lesion candidate detection and rejection to an input volume. For each detection stage an internal control mechanism ensures maintaining sensitivity on unseen true positive lesion candidates during training. This way a pre-defined target sensitivity score of the overall system can be taken into account at the time of model generation. For a lesion not only the center is detected but also, during post-processing, its spatial extension along the three spatial axes defined by the surrounding vertebral body's local coordinate system. Our method achieves a cross-validated sensitivity score of 75% and a mean false positive rate of 3.0 per volume on a data collection consisting of 34 patients with 105 osteolytic spinal bone lesions. The median sensitivity score is 86% at 2.0 false positives per volume.

Mixed spine metastasis detection through positron emission tomography/computed tomography synthesis and multiclassifier

“…Additionally, compared to prior CT-based detection systems designed for sclerotic lesion detection, 9,13,14 lytic lesion detection, 6,7 and sclerotic/lytic but not mixed type detection, 15,16 this system uses a triple classifier process to detect all three lesion types: lytic, sclerotic, and mixed. Additionally, compared to Ref.…”

“…Potential lytic bone lesions were detected using a watershed algorithm. Wels et al 7 presented a multistage lytic spinal lesion detection on CT by applying multiple random forests while maintaining sensitivity at each stage. Online feature filtering was adopted to select the most efficient features at each stage.…”

Pattern Recognition in Musculoskeletal Imaging Using Artificial Intelligence

“…Multiple computer-aided detection (CADe) software systems have been developed for the automated detection of spinal metastases. 8,[23][24][25][26][27][28][29][30] There are many systems for detecting a single type of lesion along the radiodensity spectrum, such as lytic [26][27][28] or sclerotic, [23][24][25] although some detect both types 8,29 as well as mixed lesions. 30 Among systems for the identification of osteolytic spinal lesions, the CADe algorithm developed by O'Connor et al identifies lytic lesions on CT of the chest, abdomen, and/or pelvis performed for other clinical indications.…”

“…Wels et al presented a fully automated method for the detection of osteolytic spinal lesions from three-dimensional CT entirely based on ML, except for the postprocessing step, and a multistage approach with multiple discriminative models (multiple random forests). 28 This technique resulted in a cross-validated sensitivity (mean false-positive rate per volume) of 75% (3.0).…”

Spinal Metastases Segmentation in MR Imaging using Deep Convolutional Neural Networks

“…Wiese et al [22] likewise presented an approach for sclerotic spinal metastases detection in CT images with SVMs, but modified this method with graph-cut merging of 2D regions. Wels et al [21] and Hammon et al [3] proposed different approaches using multiple random forests discriminative models. Roth et al [15] presented the first framework based on a deep CNN, while they used it as a second layer in a two-layered cascade framework to spot candidate lesions for sclerotic spine metastases detection in CT imaging.…”