A Unified Framework for Joint Segmentation, Nonrigid Registration and Tumor Detection: Application to MR-Guided Radiotherapy

Abstract

Image guided external beam radiation therapy (EBRT) for the treatment of cancer enables accurate placement of radiation dose to the cancerous region. However, the deformation of soft tissue during the course of treatment, such as in cervical cancer, presents significant challenges. Furthermore, the presence of pathologies such as tumors may violate registration constraints and cause registration errors. In this paper, we present a unified MAP framework that performs automatic segmentation, nonrigid registration and tumor detection simultaneously. It can generate a tumor probability map while progressively identifing the boundary of an organ of interest based on the achieved transformation. We demonstrate the approach on a set of 30 T2-weighted MR images, and the results show that the approach performs better than similar methods which separate the registration and segmentation problems. In addition, the detection result generated by the proposed method has a high agreement with the manual delineation by a qualified clinician.

Simultaneous Nonrigid Registration, Segmentation, and Tumor Detection in MRI Guided Cervical Cancer Radiation Therapy

“…The segmentation module here is similar to our previous work [12], [28]. In this paper, we use (2) to segment the normal or noncancerous organs (bladder and uterus).…”

“…Since a standard approach based on MR image intensity matching only may not be sufficient to overcome these limiting factors, in this paper, we present a novel probability based technique as an extension to our previous work in [28], for the motivations mentioned above. Our model is based on a maximum a posteriori (MAP) framework which can achieve deformable segmentation, nonrigid registration and tumor detection simultaneously.…”

An automatic tumor segmentation framework of cervical cancer in T2-weighted and diffusion weighted magnetic resonance images

“…To the best of our knowledge, the study about solving the problem of segmentation of cervical cancer in MR images is very limited. So far only Lu et al proposed a model in T2-weighed MR images [4], [5]. The model is based on a Maximum a Posteriori (MAP) framework which can achieve deformable segmentation, nonrigid registration and tumor detection simultaneously.…”

“…But their method only used T2-weighed MR images, and is not automatic. In their method Gross Tumor Volume (GTV), bladder, and uterus were manually segmented for therapy planning of radiotherapy in the T2-weighed MR images of each patients at first phase [5].…”

“…However, the automatic segmentation of cervical cancer in MR images is still a challenging problem [4], [5]. The structure of human abdomen is very complex, and there is noise and intensity overlapping, so the images in a single imaging modality T2-weighted MR images can not sufficiently show the precise information of the cervical cancer.…”