Published June 24, 2025 | Version v1

Machine Learning Code for Measurement of Kyphosis and Lordosis in DXA Scans

Authors/Creators

  • 1. Calico Life Sciences LLC.

Description

Measurement of Kyphosis and Lordosis from Dexa Scans

Overview

This tool applies ML models to the analysis of DEXA images for measuring curvature at the upper (thoracic) spine (kyphosis), or bottom (lumbar) spine (lordosis).

Installation

The recommended build environment for the code is to have Anaconda installed and then to create a conda environment for python 3 as shown below:

conda create -n kypho python=3.7
 

Once created, activate the environment and install all the needed libraries as follows:

conda activate kypho
pip install -r requirements.txt
 

Usage

An example for a recommended invokation of the code, to measure kyphosis:

python spineCurve.py -i <dir of imgs> -o <out file> --aug_flip --aug_tilt 0.5

...or to measure lordosis:

python spineCurve.py -i <dir of imgs> -o <out file> --aug_flip --aug_tilt 0.5 --lumbar
 
 

 

Files

KyphoLordoDxa-master.zip

Files (165.3 MB)

Name Size Download all
md5:e8530df1281be6f0add674c953167c40
165.3 MB Preview Download