Computer-aided Diagnosis of Pulmonary Embolism in Opacified CT Images

Computer-aided Diagnosis of Pulmonary Embolism in Opacified CT ImagesPh.D. Defense Presentation by Raphaël SebbeTCTS Lab — Faculté Polytechnique de Mons LESI Lab — Polytech’OrléansPlanIntroduction • Segmentation • Detection • Results • Conclusion• Context Introduction• Methods‣ Segmentation of arteries‣ Blood Clot detection‣ Assessment of performances• Results• Conclusion2ContextIntroduction • Segmentation • Detection • Results • Conclusion• Pulmonary Embolism ‣ Definition‣ Symptoms‣ Diagnosis• Medical Imaging‣ CT Scanner‣ Computer Aid ?3ContextIntroduction • Segmentation • Detection • Results • ConclusionX-ray sourceDetector4AxialCombinedCoronalSagital5Blood ClotsIntroduction • Segmentation • Detection • Results • Conclusion• Image Acquisition Protocol‣ Contrast product is used‣ A blood clot is a dark spot• Different clot types‣ Acute vs. chronic 543212345LeftRightlobarlobarsegmentalsegmentalsubsegmentalsubsegmentalTruncular6Blood ClotsIntroduction • Segmentation • Detection • Results • ConclusionClotsClots7Project GoalsIntroduction • Segmentation • Detection • Results • Conclusion• Computer Aid to Radiologists‣ Preliminary detection step‣ Increase time-efficiency of exams• Remark‣ Does not decide for the expert8State of the ArtIntroduction • Segmentation • Detection • Results • Conclusion• New problem, few solutions• Masutani et al. 2000-2003‣ Segmentation + detection‣ Region growing algorithm• Liang et al. 2005‣ Tobogganning9ApproachIntroduction • Segmentation • Detection • Results • Conclusion• Methods‣ Segmentation pulmonary arteries‣ Analysis/detection of blood clots‣ Evaluation of performances• Constrains‣ Limited computing time‣ Usability, presentation of results10