Medical Image Segmentation: How to integrate object appearance/shape models and interactive correction with minimum user intervention?

• Grant number: 15/09446-7
• Support Opportunities: Scholarships in Brazil - Post-Doctoral
• Start date: November 01, 2015
• End date: June 11, 2017
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computing Methodologies and Techniques
• Principal Investigator: Alexandre Xavier Falcão
• Grantee: Thiago Vallin Spina
• Host Institution: Instituto de Computação (IC). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
• Associated scholarship(s): 16/11853-2 - SAMSAM: Segmentation for Analysis and Measurements in the Shoot Apical Meristem, BE.EP.PD

Abstract

Problem: Medical image segmentation (in MRI and CT modalities) aims to delineate the human body's anatomical structures in a precise way to allow quantitative analysis in the diagnosis and treatment of diseases. Appearance/shape models of theses structures may be learned and applied to segment new images automatically. However, when automatic segmentation fails these methods rarely provide effective and efficient solutions to allow the interactive correction of segmentation errors without affecting parts already accepted as correct, and that minimize the user's involvement. In practice, manual/interactive segmentation is done from scratch quite often - a tedious task subject to user mistakes due to weariness and carelessness.Proposal: This postdoctoral research project aims to investigate solutions that integrate in an effective and efficiente way automatic medical image segmentation based on object appearance/shape models and interactive correction. In this context, the project aims to continue and integrate preliminary works in our group that: (a) convert \emph{any} initial segmentation into an optimum-path forest - an acyclic graph where every object of interest is represented by the union of trees rooted at nodes (seeds) inside them; (b) allow the differential correction of segmentation by the addition and removal of markers (connected seed sets) - i.e., the addition and removal of forest trees; and (c) correct minor flaws in segmentation through the smoothing of the object's boundaries. This proposal includes the development and validation of new solutions to item (a) and the validation of the current solutions to items (a) and (c). The interactive correction of automatic segmentation also allows the update of the object appearance/shape models through the inclusion of a new training sample. This aspect will also be assessed in this work.Justification and expected results: The project will be partially developed in Brazil and at the University of Sheffield, UK, under the supervision of Prof. Dr. Alejandro Frangi, thereby increasing the candidate's international experience and our group's international cooperation network. While abroad, the candidate will implement the developed methods in a public domain software (www.gimias.org), further validating the methods in images annotated by experts to increase the impact of the work beyond the expected publications.

Scientific publications

(5)

(References retrieved automatically from Web of Science and SciELO through information on FAPESP grants and their corresponding numbers as mentioned in the publications by the authors)

STEGMAIER, JOHANNES; SPINA, THIAGO V.; FALCAO, ALEXANDRE X.; BARTSCHAT, ANDREAS; MIKUT, RALF; MEYEROWITZ, ELLIOT; CUNHA, ALEXANDRE; IEEE. CELL SEGMENTATION IN 3D CONFOCAL IMAGES USING SUPERVOXEL MERGE-FORESTS WITH CNN-BASED HYPOTHESIS SELECTION. 2018 IEEE 15TH INTERNATIONAL SYMPOSIUM ON BIOMEDICAL IMAGING (ISBI 2018), v. N/A, p. 5-pg., 2018-01-01. (15/09446-7, 16/11853-2, 14/12236-1)

SPINA, THIAGO V.; STEGMAIER, JOHANNES; FALCAO, ALEXANDRE X.; MEYEROWITZ, ELLIOT; CUNHA, ALEXANDRE; IEEE. SEGMENT3D: A WEB-BASED APPLICATION FOR COLLABORATIVE SEGMENTATION OF 3D IMAGES USED IN THE SHOOT APICAL MERISTEM. 2018 IEEE 15TH INTERNATIONAL SYMPOSIUM ON BIOMEDICAL IMAGING (ISBI 2018), v. N/A, p. 5-pg., 2018-01-01. (15/09446-7)

MARTINS, SAMUEL BOTTER; SPINA, THIAGO VALLIN; YASUDA, CLARISSA; FALCAO, ALEXANDRE X.; STYNER, MA; ANGELINI, ED. A Multi-Object Statistical Atlas Adaptive for Deformable Registration Errors in Anomalous Medical Image Segmentation. MEDICAL IMAGING 2017: IMAGE PROCESSING, v. 10133, p. 8-pg., 2017-01-01. (15/09446-7, 14/12236-1, 13/07559-3, 16/11853-2)

SPINA, THIAGO VALLIN; MARTINS, SAMUEL BOTTER; FALCAO, ALEXANDRE XAVIER; IEEE. Interactive Medical Image Segmentation by Statistical Seed Models. 2016 29TH SIBGRAPI CONFERENCE ON GRAPHICS, PATTERNS AND IMAGES (SIBGRAPI), v. N/A, p. 8-pg., 2016-01-01. (15/09446-7)

MOREIRA TAVARES, ANDERSON CARLOS; VECHIATTO MIRANDA, PAULO ANDRE; SPINA, THIAGO VALLIN; FALCAO, ALEXANDRE XAVIER; ANGULO, J; VELASCOFORERO, S; MEYER, F. A Supervoxel-Based Solution to Resume Segmentation for Interactive Correction by Differential Image-Foresting Transforms. MATHEMATICAL MORPHOLOGY AND ITS APPLICATIONS TO SIGNAL AND IMAGE PROCESSING (ISMM 2017), v. 10225, p. 12-pg., 2017-01-01. (16/11853-2, 15/09446-7, 11/50761-2, 14/12236-1)