DICOM Viewer

Problem statement (hypothetical) - DICOM Viewer using Python libraries (DICOMV)

Dr. Priti is Oncoradiologist. She has around 20 years of experience analysing MRI scans of patients. She needs a custom DICOM viewer based on Python libraries.
She also wants to know how Al/ML can help her to better predict the analysis and forecast of the MRI and pathlab reports of the patients. The task is to start directly prototyping the Python DIACOM library and show how the MRI SCANs are viewed.

Please share any type of resources such as GitHub repositories, tutorial links, dataset suggestions etc.

@Pratiksha19 honestly this is a very complicated project you are talking about/proposing.

But upfront the Tensorflow library allows direct loading/decoding of DICOM files:


But to be actually be able to do anything with them, you have to understand how to build a DL/ML model and how to use Tensorflow.

Being a complete beginner how long do you think it’ll take to implement this project? Any kind of input and resources are welcomed :slight_smile:

What do you mean by this statement, seems like Dr. Priti is your primary stakeholder, wanting you to create a model for detection of diagnostic features related to MRI data available to her.

So are you seeking guidance for a work project assigned to you or something else?

I sincerely suggest you to be honest with your stakeholder while handling such projects as patient’s health are involved, be honest with your stakeholder about how much you know and don’t know, so a proper team can be formed to get a holistic approach towards a common goal of creating model that will help your stakeholder.

You might not like my response but you will not regret if you get a team of data scientist or Ai/ML engineers who have had done such projects previously where you will learn more.

DICOM VIEWER requires you to have enough images that should include all the diagnostic features your stakeholder is looking into (to understand this, you would again require to understand the anatomy and pathology of the condition you are working on), and then would come the application of findings using these images to classify or detect from the MRI images at hand you have.

first getting the images into a common format, than normalising or reshaping based on the data you have. what features would signify a finding to be considered it as bengin or malignant based on TNM stage and then the complexity of cardiology also need to be understood before proceeding to make any model.


Hey, thank you for your response. The problem statement is hypothetical and is for academic reasons.

I have noted your opinion on the same and will try to implement it.

You should have mentioned this in the post creator topic :slight_smile: