Anatomy Recognition Realized

Anatomy Detection in 3D Models

Our Partner:

With expertise in the medical device field and VC-backing, our partner was poised for success, with one setback-- the need for imaging software. Specifically, our partner needed image recognition software for anatomy detection, one of Digital Element's key strengths.

As soon as our partner and Digital Element entered a discussion about the strengths of each company it was clear that collaborating made perfect sense.

The Challenge:

Our partner needed software to detect the anatomy of patients from their 3D scans.

They needed a system where the detection algorithms worked despite: (1) some smoothing of the 3D models due to the scanning technique and (2) problems posed by the high variance in body type between different patients.

Digital Element also understood the challenges that come with being a VC-backed company. Our partner needed visible and regular progress as well as implementation of their filed patents.

In addition, since one of our partner’s principles is a scientist with multiple PhD’s, we needed work to proceed with enough focus that he could be involved in every step of the process and ultimately ‘own’ the work.

The Solution:

Digital Element helped our partner determine what milestones were reasonable within a given timeline and we made sure work was sorted and discussed in a way that kept our partner involved in every step of the process. We proceeded concurrently with the development of their hardware, documenting and communicating our work so that our partner kept control over the direction of development.

We not only created a series of algorithms that performed the required detections, but also built a development and test environment that allowed for concurrent development (inputs from other devices were used for the initial detection). A series of test cases were identified that defined the minimal, as well as optimal, levels of detection robustness.

Within the first two months Digital Element received demonstrable results, and after an effort of sixteen engineering months (4 calendar months), we robustly achieved the intended results.

The final solution included volume detection, body-feature detection, calculating the angles of visible hard tissue, auto-rotoscoping, and complex methods to effectively store data points.

The Results:

By harnessing Digital Element's imaging expertise and efficient development approach, our partner fully realized and implemented their vision-- reducing time and cost along the way.