Deep Learning in Medicine resources & Study Group

I have heard people doing 1D convolution on the chart raw data, but not the image.

1 Like

Hey @PegasusWithoutWinds and @pierreguillou does the shared google drive exist? If it does, please provide a link to it

If there is no link, then sadly this probably does not exist. Sad…

I started working on mitosis detection in breast histopathology Image using MITOS dataset. How should I make progress? Any help will be appreciable.

You can threat ECG Signal like an audio one: there is some post on the deep learning with audio thread :wink:

Deep Learning with Audio Thread?

Deep Learning with Audio Thread?

I did a bit of ECG interpretation work a few months ago, using resnet

@ste I’m interested in that audio approach.

I’m working on something else now. Trying to correlate the success of the endoscopy to the description of the procedure.

Having trouble understand exactly what my model is predicting.

NLP Question.

I’ve got a pretty simple model trying to correlate a short sentence which has been written about an endoscopy to a label which indicates whether the endoscopy was successful at visualizing the whole colon or not.

nbviewer.jupyter.org

Jupyter Notebook Viewer

Check out this Jupyter notebook!

**When my model makes a prediction, sometimes it takes the same input (e.g the terminal ileum) and makes different predictions **

I really can’t make sense of that output. Can someone with NLP experience help me out here?

1 Like

I added the kaggle retinopathy competitions to the wiki. There was one 4 years ago and the current one is finishing in 3 weeks. Very interesting problem!

I wish we had better tools to classify this type of image. Transfer learning using models that were trained on cats and dogs just doesn’t seem ideal!

1 Like

Thanks for detailed sources but I am willing to use deep learning in statistical areas especially related to sports.

Hi,

I am working on a relatively old dataset, ISIC 2016 skin lesion classification, and I am considering to try out transfer learning using model zoo/models pretrained on medical images but not aware if there is any which I can try out. I am aware of using other datasets to increase training images, but specifically I want to try it on pretrained models on medical images. Any kinds of suggestions would be helpful.

Thanks!

This is great!

Hello guys.

I’d like to share our work, sorry for a bit of self PR))

2 Likes

Hi seb!
You did put it into the wiki. Where is that?
Regards
Rolf

By “wiki”, I meant the top post of this thread (under “competitions”), which is modifiable by everyone

Edit: added the MIMIC-CXR Database

1 Like

I have download the ecg data used in a Jan 2018 publication. The EKG is in a file with .ecg extension.
Any advice on how to open and handle these files in Colab. Does Resnet treat these as image files? I tried opening these files with a photo app on Windows and they won’t open. There is an online app for viewing ECGs but it requires a license even though it says it is free.

Just thought I’d pop back in and say this thread is fantastic. I’ve been sharing some of these articles with the doctors in Melbourne and they really appreciate it. Look forward to hearing from others trying to move forward in this vein.

Hey guys - looking to start a project to use low dose CT to high dose CT but a beginner with using DICOMs and implementing them into code. Is there a step by step guide to do this? Thanks!

These are the resources that helped me when I was a DICOM beginner:

1 Like

Hi,
I’m an emergency physician and researcher in Northern California, hoping to apply ML tools to predicting sepsis. But regardless of your particular health+ML interest, if you ever need to chat to a physician about domain knowledge, just get in touch!

5 Likes
1 Like

Hello!

Michael Woodburn here, now working as a Doctor in Australia. Hoping to contribute to the ML field by creating the best dataset possible from our radiological images. I have access to 100,000 CT Abdominal Images, their metadata (name, age, etc), and freetext Radiologist-written reports. To make this dataset useful I am going to label the whole thing via various methods (automated, and human).

All reports look something like this:

“”"
CLINICAL HISTORY: ABDOMINAL PAIN.
TECHNIQUE: Multiple axial CT images were obtained through the abdomen and pelvis after
administration of oral contrast material only.
COMMENTS:
There is evidence of diffuse hepatic hypoattenuation compatible with fatty infiltration. There is no
intra or extrahepatic biliary ductal dilatation. The patient is status post cholecystectomy. The spleen
is normal. The pancreas is of normal contour and attenuation characteristics. There is no evidence of
adrenal mass.
Moderate sized fat containing supraumbilical hernia is present.
The kidneys are normal in size, shape and configuration. No renal or ureteral calculi are identified.
There is no hydroureter or hydronephrosis.
There is no evidence for appendicitis. Several fluid-filled loops of small bowel are present compatible
with mild enteritis. There is no bowel wall thickening. No evidence for small or large bowel
obstruction. There is no evidence of abdominal ascites or lymphadenopathy.
There is no evidence of intrinsic or extrinsic bladder mass. There is no pelvic ascites or
lymphadenopathy.
The uterus and ovaries are grossly unremarkable.
Images of the lung bases show no evidence of pleural or parenchymal mass. There are no pleural
effusions. Scarring is present in the right middle lobe and lingula as well as both lung bases.
The bony structures are free of lytic or blastic lesions. Multilevel degenerative changes are seen
involving the thoracolumbar spine.
Scattered calcifications are seen involving the aorta and major branches compatible with
atherosclerosis. IMPRESSION:
ABDOMEN:

  1. Several fluid-filled loops of small bowel are present compatible with mild enteritis.
  2. Fatty liver.
  3. Status post cholecystectomy.
  4. Fat containing umbilical hernia.
    PELVIS:
  5. No evidence of diverticulitis or acute inflammatory process in the pelvis.
    “”"

And we have some idea of the specific labels we will derive from these reports. For example:

  1. Pathologies (diseases, e.g appendicitis)
  2. Radiological features (e.g fat stranding)
  3. Report written by trainee vs senior
  4. Follow up study required
  5. Scan of sufficient quality to report

My big question to you all is
Can anyone think of any other interesting labels/targets we should be looking for?
Something a little left of field perhaps?

Open to any and all suggestions. Looking forward to making this dataset available to interested ML enthusiasts in future.

2 Likes