Improved Computer-aided Detection and Annotation of Spiculated Masses in Mammography

[UT Grad StudN]

Hi, I'm a graduate student at the University of Texas, I'm currently working with a professor who I believe has made significant developments in the ability to detect spiculated masses with CADe software. I'm trying to determine if their is an appetite in the radiologist community for her new software.

Her patent-protected technology, implemented through software, provides a model-based, evidence-based approach to capture the exact shape of spicules on mammograms using a modular, flexible image analysis technique. The system can be easily integrated into existing CADe systems and offers an intuitive depiction of the spicules on the mammogram to alert the radiologist. The invention uses a series of Gaussian filters, spiculated lesion filters and Difference of Gaussian filters to produce a combined enhanced image.

I'm trying to determine the answer to the following questions
•Do you use computer-aided detection technology (CADe) to analyze mammograms?
•(If so) Which software system(s) do you use? Does CADe help or hinder you?
•Would enhanced Mammograms save you time?
•Would you like a product that provided better clarity to the mammogram?
•In your experience what is most difficult about the process of detecting breast cancer?
•What improvements would you like to see made to breast cancer detection?
•Are false positive detections of spiculated masses a large problem in your practice?

I attached a picture to demonstrate the results of her work.

Any feedback is appreciated.

Thanks,
UT Grad StudN

---

Members don't see this ad.

 

[TCOM12]

Since a lot of places are moving to 3D breast tomosynthesis, this technology may not be useful to many. The toughest thing is usually detecting microcalcifications and CAD is already implemented to detect those. If you want feedback from radiologists, I would go to a different site that's not allowed to be named here.

Good luck

 

[Radology]

Most mammographers don’t really use cad to be honest. It started as a marketing tool and then became standard because of a couple flawed studies saying that it helped someone with little breast experience read a mammogram

If it could be improved, it would be wonderful. But the example you showed is about as blatantly obvious as it gets without cad.

 

[deleted103644]

No radiologist will pay for this on top of CAD. Honestly the only reason radiologists pay for CAD is because it counts as a second reader.

If it is useful at all, an existing CAD company might license your technology, but I don’t see much use for it. And aside from the vague term “spiculated lesion filter” which could mean anything, the other things you mention are old hat in image analysis and I would assume are already built into CAD systems.

 

[UT Grad StudN]

Thanks for the responses! I think I'll shift my efforts to a different pursuit.