January 4, 2023

AI Diagnosis of Hip Misalignments proves Reliable, Fast and Cost Efficient

A validation study by The University of Texas Southwestern Medical Center in Dallas (UTSW) confirms the advantages of an artificial intelligence software by ImageBiopsy Lab as part of the standard diagnosis of hip dysplasia

Share this article

AI Diagnosis of Hip Misalignments proves Reliable, Fast and Cost Efficient

A validation study by The University of Texas Southwestern Medical Center in Dallas (UTSW) confirms the advantages of an artificial intelligence software by ImageBiopsy Lab as part of the standard diagnosis of hip dysplasia

January 4, 2023

The Deployment of an artificial intelligence (AI) based image analysis software for the diagnosis of hip dysplasia can save time and costs without compromising the reliability of the diagnosis. This is the result of a recently published external validation study that tested IB Lab HIPPO - a musculoskeletal software tool by ImageBiopsy Lab, Austria (IB Lab). Currently the diagnosis of hip dysplasia, an abnormality of the hip joint, is based on time and cost consuming manual radiological measurements.  The study conducted at the University of Texas Southwestern Medical Center shows that using an AI tool for image analysis can significantly accelerate the diagnosis and result in saving labor costs of highly trained experts. In a cohort of 256 hips, HIPPO managed to perform all six relevant measurements for Hip Dysplasia assessment and the study confirmed good to excellent inter-reader reliability for important bone parameters measured by HIPPO and trained experts, demonstrating the reliability of artificial intelligence based automated analysis. Recently the results have been presented at the annual meeting of the Radiological Society of North America (RSNA).

Approximately 5 – 10% of the population are affected by hip dysplasia, a developmental condition where the bones of the hip joint are misaligned. If not treated or treated too late, it leads to pain, instability and premature Osteoarthritis.  Several complex radiological measurements are used for assessing the extent of the joint abnormality. However, unstandardized measurements and high inter-reader variability lead to statistically and clinically relevant differences in hip diagnosis and potentially to inadequate treatment. Employing an AI based image analysis software however may contribute to more standardized and reproducible measurements – if it proves to be as accurate as the current time-consuming gold standard, manual  measurements of images by trained experts.

Reliable. Fast. Cost Efficient.

“In summary”, announces Dr. Richard Ljuhar, CEO and Co-founder of IB Lab, “the study confirms that for the vast majority of analyzed images the AI based method comes to essentially the same measures as the ones obtained by trained experts – just much faster and thus considerably cheaper.”

In detail, the team UTSW Medical Center used radiological images of 256 hips. From each image 6 measurements were taken: Lateral center-edge angle, caput-collum-diaphyseal angle, pelvic obliquity, Tönnis angle, Sharp’s angle and femoral head coverage, either by HIPPO or by three trained experts. When comparing the results obtained by either method, they showed good to excellent correlations, ranging in average from 0.6 to 0.98 (with 1 being identical results). Even better results were achieved when the clinically most widely used measurements (lateral center-edge and Tönnis angle) were compared. Here, the correlation was within 0.71 to 0.86 and 0.82 to 0.90, respectively.

Fig. 1 Flowchart for final study sample containing preoperative hip dysplasia patients with complete radiological imaging who met artificial intelligence (AI) requirements.

Saves Time & Money

In the study three human trained experts were asked to perform manual readings. The median time for reading a single image varied widely between the three individuals, from 131 seconds to 734 seconds. “This highlights a problem with the standard diagnosis approach”, explains Dr. Ljuhar. “The time required for human based analysis vastly depends on the individual and hinders efficient and coordinated workflows across institutions.” HIPPO in comparison only needed 41 seconds per image (median value) with very little variation at all, resulting in time savings between 70 and 90 % and the possibility to standardize diagnostic procedures across institutions.

Fig. 2 Landmarks used by manual readers. a) Lateral centre-edge angle. b) Caput-collum-diaphyseal angle. c) Obliquity. d) Tönnis angle. e) Sharp’s angle. f) Femoral head coverage.
Fig. 3 Example HIPPO output. This figure shows an example of the reports that HIPPO produces.

But time is not all that can be saved by using HIPPO. Based on average salaries for either orthopedic surgeons or radiologists and the aforementioned time savings, up to 80% of labor costs for assessing the radiographs can be saved. In detail, the average cost for analysis of one image by an orthopedic surgeon is 36.50 USD whereas the HIPPO supported assessment would cost an equivalent of 4.18 USD of the surgeon’s time. For a radiologist the values are 28.66 USD to 3.27 USD. In an internal calculation IB Lab extrapolated these numbers for medical institutions that analyze 10,000 radiographs per annum. “For analysis by orthopedic surgeons that totals 320,000 USD per annum”, comments Dr. Ljuhar. “Considering a license fee for such a volume of hip X-rays, that equals more than 85% annual savings potential.”

The recent study confirms IB Lab’s position at the cutting edge of AI based medical image analysis. Besides HIPPO for hip abnormalities the company offers analysis modules such as PANDA for bone age assessments, FLAMINGO for detection & quantification of silent vertebral fractures in the spine and SQUIRREL for spinal disorders. More modules are consistently developed and are assessed jointly with some of the most prestigious medical institutions world-wide.


Artificial intelligence-generated hip radiological measurements are fast and adequate for reliable assessment of hip dysplasia

an external validation study

Holden Archer, Seth Reine, Ahmed Alshaikhsalama, Joel Wells, Ajay Kohli, Louis Vazquez, Allan Hummer, Matthew D. DiFranco, Richard Ljuhar, Yin Xi, Avneesh Chhabra

Published in Bone & Joint Open, Vol. 3, No. 11: https://doi.org/10.1302/2633-1462.311.BJO-2022-0125.R1