Detail of the student project

List
Topic:Detekce potenciálních plicních nodulů z CT obrazů
Department:Katedra kybernetiky
Supervisor:prof. Dr. Ing. Jan Kybic
Announce as:Diplomová práce, Bakalářská práce, Semestrální projekt
Description:Lung nodules are small objects in lungs that might be malignant. The standard approach for nodule detection from 3D CT images consists of two steps. First a large number of candidate nodules are identified. In the second step, the nodules candidates are classified to suppress false positive detections. This work focuses on the first subproblem.

Tasks

- Familiarize yourself with existing work on nodule detection and classification. Describe the state of the art.

- Familiarize yourself with existing datasets and challenges (e.g. TIME, LIDC, ANODE09,LUNA2016…). Learn how to read and visualize the images and the detection results.

- Learn to use the existing pipeline from the Bc work of Maria Dobko.
Implement one of the classical nodule detection methods such as a blob detector. Evaluate its performance independently and when used with the existing classifier.

-[time permitting] Implement a CNN based candidate detector. Evaluate its performance independently and when used with the existing classifier.

-[time permitting] Suggest, implement and test improvements for the nodule candidate detector.
Responsible person: Petr Pošík