Project 1

Shreya D'Souza

VISUALISING BRAIN TUMOUR PROGRESSION IN RESPONSE TO CHEMOTHERAPY

Dataset

https://wiki.cancerimagingarchive.net/display/Public/Brain-Tumor-Progression

"This collection includes datasets from 20 subjects with primary newly diagnosed glioblastoma who were treated with surgery and standard concomitant chemo-radiation therapy (CRT) followed by adjuvant chemotherapy. Two MRI exams are included for each patient: within 90 days following CRT completion and at progression (determined clinically, and based on a combination of clinical performance and/or imaging findings, and punctuated by a change in treatment or intervention). "

I have found data files from within the dataset that give me quite a good visualisation of the tumour. The only possible issue is that the data is from 1991/1992, which could be considered outdated. The images are quite clear, however, and treatment for cancer has not evolved greatly since then. I am using MRIs, which give the clearest image of brain tumours.

The video on the top right is a screen recording of me visualising an MRI from the dataset

Screen Recording 2020-02-05 at 9.22.41 PM.mov

Purpose

I will be adding colour to the tumour and seeing how it progresses over time. A possible application of this would be to give surgeons a 3D-view of the brain so they can plan out surgeries and where to cut into the brain.

Useful papers outlining more about the purpose and implementation:

I think the project scope will be limited to visualising the tumour and porting it to the Yurt, but it'll be interesting to see if machine learning techniques can be applied to the data so that the way in which tumours react to chemotherapy can be predicted and we can have a better understanding of how treatment of tumours will pan out.

Goals

  1. Determine a reliable data set to use for the course of the project:

    • Data seems to come in three "views", as shown on the left

    • Combine the three views to create one 3D representation of the brain

  2. Identify and isolate tumour regions

  3. (If data permits), show progression of tumour over time

  4. Find a way to port the visualisation to the YURT

  5. Create a video showing my process

  6. Create a tutorial for downloading MRI datasets and using 3D Slicer to create 3D representations

Class Activity

  • Demonstrating 3D Slicer: There is a lot of sample data within 3D Slicer, including data of brain MRIs. I could also show my present method of visualising my data so that people can critique the quality of my visualisation.

  • Another possible class activity: Identifying reliable data for visualisation

Deliverables

  • A video showing the progression of the tumour for multiple patients. I still need to look at how the data varies across patients. The tumours will be shrinking over time, but it will be interesting to see how different patients respond to chemotherapy. I will also include how to identify reliable data for visualisation and how to use the Cancer Imaging Archive to download the relevant data files. I hope to have my project fully portable to the YURT so I can include footage of it running in the YURT.

    • This will show the potential of 3D Slicer to the Wiki users and what the constructed brain image looks like in VR

    • This could be almost like a timelapse.

  • Tutorial for 3D slicer for the Class wiki

    • I can have a writeup on the class wiki and select footage of me using 3D Slicer to accompany my writeup

  • (If time allows, not sure if this is a useful deliverable) A poster showing my process, showing how I created my visualisations.

Plan

[Everything written is what I plan to have done by the respective date]

2/11:

  • Finalise datasets by visualising data in 3D Slicer and seeing if MRI scans can be combined to produce a good 3D visualisation

  • Find out how to properly identify glioblastomas in MRIs so I properly identify them and isolate them from the MRI scans

  • Work on tutorial showing how to identify good data for visualisation.

2/13:

2/20:

  • If not completed, 3D visualisations of brain with the tumours, so they look something like the visualisation below. Ensure that multiple scans for the same patient are chosen so the progress can be seen.

  • Work on Yurt portability: determine if it is possible to take what cam be shown on a headset and put it in the Yurt

  • Finish plan for the class activity

2/25:

  • Work on Yurt portability, should have a working version by now.

  • Start working on progress video deliverable

    • Show how MRI scans were used to create a 3D representation and how the tumour was identified

2/27:

  • Continue working on Yurt portability

  • Visualise at the YURT

  • Take videos of how it looks at the YURT for the video deliverable

3/03:

  • Project should be fully portable to the Yurt, should have multiple visualizations

  • Video showing the process should be done

  • Writeup accompanying the video should be done