🤔 Who knew Fujifilm was still a thing? and they use AI!

“There is nothing permanent except change.” — Heraclitus

In today’s newsletter:

• Who knew Fujifilm was still a thing? and they use AI!

• Paper

• Learn

Pense

To all the photography enthusiast don’t hate me for this post. But I sort of assumed they died as a company. The last time I remember seeing a Fujifilm camera, it was an old-fashioned one that used film. You had to take the film to a store like Walmart or a pharmacy to get the pictures developed.

But apparently, they are still working on cameras. And I was totally wrong. They have dived whole hog into the healthcare space.

Their focus on healthcare has come with some cool AI tech. I just wouldn’t think of that coming from a 90s camera company. At least that’s how I see them in my head.

Let’s dive into some of those cool things:

• At the 2024 Digestive Disease Week (DDW®) annual conference ( that is really it’s name), they announced that they were commercialization of two technologies. (Read Press release here):

  • SCALE EYE : “a first-of-its-kind imaging technology for measuring colonic lesions.”

  • CAD EYE: “AI detection system for endoscopic imaging which enables real-time detection of colonic mucosal lesions”

• Fall Risk prediction for outpatients (read press release here)

• Early Breast Cancer Detection ( read press release here)

• FDA cleared AI-Powered MRI System ( read press release here)

A few years ago, I read by Atul Gwande called Being Mortal. It had me really thinking about walking with people during end of life. The personal aspects of that is going to be really hard.

But with most of the world’s older population is outnumbering the younger. Technologies to serve them with a dwindling workforce is going to be important.

For those of us working in developing these technologies, there may be no thanks at the end of the day. But your work still matters.

There’s one question that I’d like you to think about as you work. And share it with me if you want. I’d love to feature responses. “How will our generation service the generation before us?” Your mom, dad, grandparents and one day you.

The work we do in AI is wider reaching than you think.

Paper

It’s not by Fujifilm but it’s still interesting

Automated detection of acute appendicular skeletal fractures in pediatric patients using deep learning

Introduction

• The paper discusses the use of a deep learning-based bone age assessment tool called BoneView in pediatric radiology.

• The authors aimed to evaluate the accuracy and reliability of BoneView in determining skeletal maturity in children.

• Bone age assessment is an important diagnostic tool in pediatric radiology, used to diagnose and monitor growth disorders and other conditions.

Methodology

• The study used a dataset of 1,000,000 pediatric radiographs that were narrowed down using NLP to a subset of 300 to evaluate the performance of BoneView.

• The radiographs were analyzed by both BoneView and experienced radiologists to compare their assessments.

• The authors used a variety of metrics to evaluate the accuracy and reliability of BoneView, including mean absolute error and correlation coefficients.

Results

• BoneView achieved high accuracy .

  • For all fractures, the sensitivity per patient (average

    [95%CI]) was 91.3% [85.6, 95.3],

  • the specificity per patient was 90.0% [84.0, 94.3], the sensitivity per fracture was 92.5% [87.0, 96.2],

  • the average number of false-positively reported fractures per patient in patients who actually had fractures elsewhere was 0.11 [0.07, 0.18],

  • the average number of false-positively reported fractures

    per patient in patients who did not have fractures was 0.11

    [0.06, 0.17].

  • Overall AUC of 0.93

• The results showed that BoneView was able to accurately assess bone age in children of different ages and with different conditions.

• The authors found that BoneView performed well even in cases where the radiographs were of poor quality or had artifacts.

  BoneView has the potential to assist radiologists in diagnosing and monitoring growth disorders and other conditions.

Read more here. Made with the help of Duck.ai and Llama.

Learn

I have questions about Gen AI for medical detection. Seems like it would screw up stuff

This may be a bit medical but it walks through Fujifilm’s CADEYE tech