DL4H Alzheimers CNN

[sgolara2]

Contributor name and NetID/email: Soheil Golara (sgolara2) and Karan Desai (karansd2)
Type of contribution (dataset+task/model/task): dataset+task
Link to original paper (required for reproducibility contributions): https://arxiv.org/abs/1911.03740
High-level description of implementation: We have implemented a new dataset for MRI Brain Scans from an opensource OASIS repository. The dataset includes NIFTI images with and without dementia and we've added a new task+processor to support that new file type.
File guide listing which files to review:
docs/api

datasets
    pyhealth.datasets.MRIDataset.rst
tasks
    pyhealth.tasks.MRIBinaryClassification.rst
datasets.rst
tasks.rst

examples
pyhealth

datasets
    configs
        mri.yaml
    init.py
    mri_dataset.py
models
    cnn.py
    embedding.py
processors
    init.py
    nifti_image_processor.py
tasks
    init.py
    mri_binary_classification.py

test-resources/core/mri

oasis_cross-sectional.csv

tests/core

test_mri.py

pixi.lock
pyproject.toml
Follows example PR format from contribution guide: https://github.com/sunlabuiuc/PyHealth/pull/392/changes

[Jathurshan0330]

Good PR

[EricSchrock]

@Jathurshan0330 #921 overlaps this PR and is also on the list of high-quality PRs John sent out. Perhaps we should compare them to determine which one is best and/or take the best parts from each?