Interactive Segmentation For Image Guided Therapy Ohad Shitrit

What are we going to speak about? q Computed Tomography q Motivation q Mathematical

Computed Tomography (CT) Spiral Cone-Beam Scanning for Computed Tomography. Ge Wang, 2003 (All rights

Computed Tomography (CT) q X-Ray Projection using radon transform Spiral Cone-Beam Scanning for Computed

Computed Tomography (CT) q Radon transform as one dimensional Fourier transform q Reconstructing the

Why is there any need for interactive segmentation ? Tsachi H. & Ohad S.

Why is there any need for interactive segmentation? q Volume estimation is critical for

“Active Contour” But first things first… Tsachi H. & Ohad S.

Probabilistic Model Based on Gaussian Mixture(GM) q We will define the Probability of a

A CT scan Histogram of a brain with Cerebral hemorrhage Tsachi H. & Ohad

Level Set Function - Tsachi H. & Ohad S.

Mathematical Issues Tsachi H. & Ohad S. [Riklin Raviv, Van Leemput, Menze, Wells, Golland,

Problem Definition To achieve the optimized segmentation we maximize the joint distribution: Tsachi. H.

Energy Functional q Using the following relationship: q Allows us to formulate our problem

Image Likelihood Term q Assuming Gaussian Mixtures Model (GMM) Tsachi H. & Ohad S.

Smoothness Term q Objects in nature are continuous q Trade off between smoothness and

User Interaction Term Tsachi H. & Ohad S.

Gradient Descent The gradient descent is an iterative process which leads to the minimum

Block Diagram – Entire System Tsachi H. & Ohad S. [Riklin Raviv, Van Leemput,

Performance Dice Sensitivity Specificity Accuracy Automatic 0. 874± 0. 034 0. 864± 0. 073

Data q Provided by Dr. Ilan Shelef, Department of Radiology, Soroka University Medical Center

Conclusion q Semi-automatic segmentation tool q Probabilistic model q User Interface q Collaboration with

Future work q Adjustments to other modalities (MRI) q Handle with various of structures

User Interface VIDEO 4 Tsachi H. & Ohad S.

Questio ns? Demo - http: //youtu. be/Jb-6 VDid 37 s

Slides: 44

Download presentation

Interactive Segmentation For Image Guided Therapy Ohad Shitrit & Tsachi Hershkovich Ben-Gurion University of the Negev

What are we going to speak about? q Computed Tomography q Motivation q Mathematical introduction q Problem definition q Energy q Gradient Descent q The system q Results q Conclusion & Future work Tsachi H. & Ohad S.

Computed Tomography (CT) q Radon transform as one dimensional Fourier transform q Reconstructing the image with the inverse Fourier transform Tsachi H. & Ohad S.

Why is there any need for interactive segmentation ? Tsachi H. & Ohad S.

VIDEO 1 Tsachi H. & Ohad S.

Why is there any need for interactive segmentation? q Volume estimation is critical for further treatment q Therapist knowledge is essential for final decisions q Fast and accurate analysis might save life Tsachi H. & Ohad S.

VIDEO 2 Tsachi H. & Ohad S.

VIDEO 3 Tsachi H. & Ohad S.

“Active Contour” But first things first… Tsachi H. & Ohad S.

Probabilistic Model Based on Gaussian Mixture(GM) q We will define the Probability of a Voxel (3 D pixel) to belong to the Object Or to the Background: Tsachi H. & Ohad S.

A CT scan Histogram of a brain with Cerebral hemorrhage Tsachi H. & Ohad S.

Level Set Function - Tsachi H. & Ohad S.

Mathematical Issues Tsachi H. & Ohad S.

Mathematical Issues Tsachi H. & Ohad S. [Riklin Raviv, Van Leemput, Menze, Wells, Golland, Medical Image Analysis, 2011]

Problem Definition To achieve the optimized segmentation we maximize the joint distribution: Tsachi. H. H. &&Ohad. S. S. [Riklin Raviv, Van Leemput, Menze, Wells, Golland, Medical Image Analysis, 2011]

Energy Functional q Using the following relationship: q Allows us to formulate our problem as an energy minimization problem q Summing all contributions from each voxel Tsachi H. & Ohad S.

Image Likelihood Term q Assuming Gaussian Mixtures Model (GMM) Tsachi H. & Ohad S.

Smoothness Term q Objects in nature are continuous q Trade off between smoothness and sensitivity q Fine tuning is needed Tsachi H. & Ohad S.

User Interaction Term Tsachi H. & Ohad S.

Gradient Descent The gradient descent is an iterative process which leads to the minimum of the Energy term. Tsachi H. & Ohad S.

Block Diagram – Entire System Tsachi H. & Ohad S. [Riklin Raviv, Van Leemput, Menze, Wells, Golland, Medical Image Analysis, 2011]

Entire System 3 D

Entire System 3 D Tsachi H. & Ohad S.

Performance Dice Sensitivity Specificity Accuracy Automatic 0. 874± 0. 034 0. 864± 0. 073 0. 998± 0. 0019 0. 996± 0. 0033 With user interaction 0. 905± 0. 027 0. 870± 0. 063 0. 999± 0. 0003 0. 997± 0. 0022 Tsachi H. & Ohad S.

Data q Provided by Dr. Ilan Shelef, Department of Radiology, Soroka University Medical Center and Zlotowski Center for Neuroscience, Ben-Gurion University of the Negev. q Modality: CT Brilliance 64 q Resolution: 512 x. Z (Z = 90 -100) X x Y x Z = 0. 48 x 3 [mm] q Z axis with 1. 5[mm] overlap Tsachi H. & Ohad S.

Conclusion q Semi-automatic segmentation tool q Probabilistic model q User Interface q Collaboration with Soroka Medical Center Tsachi H. & Ohad S.

Future work q Adjustments to other modalities (MRI) q Handle with various of structures q User-Machine dialog in the medical world Tsachi H. & Ohad S.

User Interface VIDEO 4 Tsachi H. & Ohad S.

Questio ns? Demo - http: //youtu. be/Jb-6 VDid 37 s