Left Ventricle Segment.pptx

Segmenting

1. Image Segmentation Paper# title: Left ventricle segmentation

2. Introduction Early diagnosis of CVDs often time-consuming. Need to automate cardiac structure segmentation for image analysis LV Segmentation Challenging problem : LV confused with other MRI regions Expert own experience

3. Research Objective A ddress LV segmentation in cardiac MRI Build a model locate the ROI segment the LV region Combined CNN and U-Net model

4. Performance Evaluation Evaluation metrics used : Volumetric Overlap Error Prediction error rate Hausdorff Distance Distance metrics Dice Metrics Overlapping effect

5. Limitations Inconsistent segmentation boundaries Combined model how it reduces segmentation inaccuracies Insufficient data for model training N ot possible to evaluate model portability.

6. Conclusion The researchers address the problem of LV segmentation in cardiac MRI. Built a composite model which combines the CNN and U-net. CNN locates the center position of LV crops the ROI from the original MRI U-net completes the segmentation of LV in the ROI

7. Critical Analysis Segmentation with single or combined CNNs – or modified versions U-Net LV segmentation Localize borders, pixels Touching objects Segmentation mask Fewer training samples

8. Critical Analysis Combined CNN and U-Net model : accurate LV segmentation . CNN: Locate ROI U-Net: LV segmentation Seg mentation based of viewpoints Semantic segmentation

9. Thank you!