ohbm / hackathon2019

Website and projects for the OHBM Hackathon in Rome 2019
https://ohbm.github.io/hackathon2019
83 stars 59 forks source link

Automated Cortical Lesion Detection using Python Tools #66

Open barbrakr opened 5 years ago

barbrakr commented 5 years ago

Cortical Lesion Finder

Barbara A.K. Kreilkamp

Project Description

The high anatomic specificity of MRI may depict focal lesions and can be expertly assessed by visual analysis through neuroradiologists (Von Oertzen et al. 2002). Still, it is important to find ways to improve the diagnostic yield from MRI through optimized MRI protocols, expert neuroradiological assessment and quantitative analysis of post-processed volumetric MRI (Sisodiya et al. 1995, Huppertz et al. 2005).

This project focuses on quantitative analysis to improve detection of focal cortical dysplasia (FCD), which is a common lesion associated with medically refractory epilepsy and often epileptogenic. FCD is a type of cortical malformation that is neuroradiologically characterized by cortical thickening, GM/WM blurring and transmantle signs, which are abnormal extensions of GM towards the ventricles (Barkovich et al. 1997, Huppertz et al. 2005). FCDs are the most common lesions in children and is the third most common lesion after hippocampal sclerosis (HS) and tumors in adult patients.

Within our study, a dedicated epilepsy MRI research protocol including isotropic 3D T1-weighted and FLAIR was performed on patients with medically refractory focal epilepsy, who were deemed to be non-lesional based on previous MRI. The most recent MRIs conducted in context of this study allowed (i) a clinical diagnostic assessment by an experienced neuroradiologist and (ii) the application of an automated quantitative voxel-based lesion detection technique on patients' MRIs in order to find potentially epileptogenic lesions such as FCDs.

I have used MATLAB to program an automatic cortical lesion finder tool and would like to translate it into Python together with you!

image

Skills required to participate

Experience in Python (and possibly MATLAB, not a requirement) Creativity for incorporating SPM12, nipype and nilearn (for voxel-based morphometry)

Integration

As of now, we only have a limited number of cortical lesions. The idea is to make this project available to clinicians as collaborators, incorporate their feedback and improve the detection rate and usability of the software.

Milestones:

(i) Design a user-friendly and low-level Graphical User Interface; (ii) read in MRI data (nifti or preferred DICOM); (iii) translate MATLAB/SPM12 algorithms using GitHub packages

Preparation material

Join Mattermost channel

@cortical_lesion_finder

GitHub Repository

Cortical Lesion Finder

More References

barbrakr commented 5 years ago

Hi there, Would you please help me label this as a OHBM hackathon project? Thanks & kind regards, Barbara

emdupre commented 5 years ago

Just added the label ! Thanks for proposing this project -- it looks really interesting !!

barbrakr commented 5 years ago

Thank you, Elizabeth! I came across this - visual reports project of yours, certainly something that could be great for us to do too - otherwise clinicians won't know what's going on within the software while they're running it.

emdupre commented 5 years ago

Hey @barbrakr ! Where should folks who are interested in working on this meet you ??

CarstenSchmidt-Samoa commented 5 years ago

We are at one of the higher tables close to the projection screen on level 2.

barbrakr commented 5 years ago

Please check where we are on the mattermost channel - thanks!