MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel

A Machine Learning Based Fully Automatic Magnification Calculation Method for Hip DR Photography

Y Jia1*, H Wang1, X Jin2, H Du1, W Chen1, B Yang2, (1) Shaanxi Key Laboratory of Network Data Intelligent Processing; UCLA School of Medicine, Xi'an, Shaanxi, CN, (2) Xian Honghui Hospital

Presentations

(Sunday, 7/12/2020)   [Eastern Time (GMT-4)]

Room: AAPM ePoster Library

Purpose: existing magnification correction method used in our clinical practice is based on the target and an object of reference, which is moderately accurate and not reliable enough in clinical practice. The aim of this study is to develop an accurate model for X-ray magnification correction for hip X-rays to provides more accurate measurements for surgeons to optimize their surgical plans or design better mechanical devices for prosthesis replacement.


Methods: digital X-ray images of the hip were collected from the hospital. We recorded age, gender, the thickness of the body, the size of the limb, and the size of the limb in the X-ray image for the modeling of the limb to image distance (LID). We split the images and records into training and validating datasets (58:56) with no overlap. We trained a mathematical model based on least-squares method to estimate the LID with other factors, and then calculate the magnification according to the function of the magnification and LID. The magnification values measured by experts from the hospital were used as ground truth (GT), and the absolute deviation of the estimation and the GT was calculated to evaluate the model.


Results: build a model trained with clinical information and some measurements easier to estimate the LID and magnification of the digital X-ray images of hip. The absolute deviations between the estimated LID, magnification, and the GTs are 0.26±0.15 and 0.004±0.005 respectively. Professional surgeons from the hospital tested the model and claimed the accuracy of the model meets the clinical needs.


Conclusion: work indicates the potential for estimating the LID and magnification in X-ray imaging with a mathematical model. The procedure is easy to apply in clinical practice and it can provide accurate measurements for preoperative planning, surgical simulation, and prostheses design.

Keywords

Calibration, X Rays, Modeling

Taxonomy

IM- X-Ray: Digital radiography (DR and CR)

Contact Email