Segmenting Glioma in Multi-Modal Images using a Generative-Discriminative Model for Brain Lesion Segmentation

نویسندگان

  • Bjoern H. Menze
  • Ezequiel Geremia
  • Nicholas Ayache
  • Gabor Szekely
چکیده

In this paper, we evaluate a generative-discriminative approach for multi-modal tumor segmentation that builds – in its generative part – on a generative statistical model for tumor appearance in multi-dimensional images [1] by using a “latent” tumor class [2, 3], and – in its discriminative part – on a machine learning approach based on a random forest using long-range features that is capable of learning the local appearance of brain lesions in multi-dimensional images [4, 5]. The approach combines advantageous properties from both types of learning algorithms: First, it extracts tumor related image features in a robust fashion that is invariant to relative intensity changes by relying on a generative model encoding prior knowledge on expected physiology and pathophysiological changes. Second, it transforms image features extracted from the generative model – representing tumor probabilities in the different image channels – to an arbitrary image representation desired by the human interpreter through an efficient classification method that is capable of dealing with high-dimensional input data and that returns the desired class probabilities. In the following, we shortly describe the generative model from [1], and input features and additional regularization methods used similar to our earlier discriminative model from [4].

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Segmenting Glioma in Multi-Modal Images using a Generative Model for Brain Lesion Segmentation

In this paper, we evaluate a fully automated method for channel-specific tumor segmentation in multi-dimensional images proposed by us in [1]. The method represents a tumor appearance model for multi-dimensional sequences that provides channel-specific segmentation of the tumor. Its generative model shares information about the spatial location of the lesion among channels while making full use...

متن کامل

A generative probabilistic model and discriminative extensions for brain lesion segmentation – with application to tumor and stroke

We introduce a generative probabilistic model for segmentation of brain lesions in multi-dimensional images that generalizes the EM segmenter, a common approach for modelling brain images using Gaussian mixtures and a probabilistic tissue atlas that employs expectation-maximization (EM) to estimate the label map for a new image. Our model augments the probabilistic atlas of the healthy tissues ...

متن کامل

A Generative Model for Brain Tumor Segmentation in Multi-Modal Images

We introduce a generative probabilistic model for segmentation of tumors in multi-dimensional images. The model allows for different tumor boundaries in each channel, reflecting difference in tumor appearance across modalities. We augment a probabilistic atlas of healthy tissue priors with a latent atlas of the lesion and derive the estimation algorithm to extract tumor boundaries and the laten...

متن کامل

Automatic segmentation of glioma tumors from BraTS 2018 challenge dataset using a 2D U-Net network

Background: Glioma is the most common primary brain tumor, and early detection of tumors is important in the treatment planning for the patient. The precise segmentation of the tumor and intratumoral areas on the MRI by a radiologist is the first step in the diagnosis, which, in addition to the consuming time, can also receive different diagnoses from different physicians. The aim of this study...

متن کامل

Automatic Caudate Segmentation by Hybrid Generative/Discriminative Models

Segmenting sub-cortical structures from 3D brain images is of significant practical importance. This paper presents an experimental study for caudate segmentation in MRI images using a fully automatic algorithm based on [9], which is an extension of the hybrid model approach described in [1]. The method in [9] tackles multiple sub-cortical and cortical structures. In this study (the grand chall...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2012