Fully automated deep learning for knee alignment assessment in lower extremity radiographs: a cross-sectional diagnostic study

Simon, Sebastian; Schwarz, Gilbert M.; Aichmair, Alexander; Frank, Bernhard J. H.; Hummer, Allan; DiFranco, Matthew D.; Dominkus, Martin; Hofstaetter, Jochen G.

doi:10.1007/s00256-021-03948-9

Cited by 37 publications

(43 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Therefore, an increasing number of studies have used DCNNs to automatically measure these angles and lengths on plain films. Schock et al [ 18 ] used a DCNN algorithm to measure hip-knee-ankle angle and femoral anatomic-mechanical angle automatically and quantitatively and the measurements were as precise and accurate as manual reference measurements with 3–7 s. Recently, Simon et al [ 19 ] tested the LAMA software, which was trained on over 15,000 radiographs from multiple centres using DCNN to measure the hip-knee-ankle angle, anatomical mechanical angle, JLCA, mLDFA, LDTA, mechanical lateral proximal femoral angle, MPTA, mechanical-axis-deviation, leg length, femur length, and tibia length. The software achieved an overall accuracy of 89.2% when comparing the AI outputs to those that were manually measured.…”

Section: Discussionmentioning

confidence: 99%

Fully automated measurement on coronal alignment of lower limbs using deep convolutional neural networks on radiographic images

Meng

Wang

et al. 2022

BMC Musculoskelet Disord

View full text Add to dashboard Cite

Background A deep convolutional neural network (DCNN) system is proposed to measure the lower limb parameters of the mechanical lateral distal femur angle (mLDFA), medial proximal tibial angle (MPTA), lateral distal tibial angle (LDTA), joint line convergence angle (JLCA), and mechanical axis of the lower limbs. Methods Standing X-rays of 1000 patients’ lower limbs were examined for the DCNN and assigned to training, validation, and test sets. A coarse-to-fine network was employed to locate 20 key landmarks on both limbs that first recognised the regions of hip, knee, and ankle, and subsequently outputted the key points in each sub-region from a full-length X-ray. Finally, information from these key landmark locations was used to calculate the above five parameters. Results The DCNN system showed high consistency (intraclass correlation coefficient > 0.91) for all five lower limb parameters. Additionally, the mean absolute error (MAE) and root mean squared error (RMSE) of all angle predictions were lower than 3° for both the left and right limbs. The MAE of the mechanical axis of the lower limbs was 1.124 mm and 1.416 mm and the RMSE was 1.032 mm and 1.321 mm, for the right and left limbs, respectively. The measurement time of the DCNN system was 1.8 ± 1.3 s, which was significantly shorter than that of experienced radiologists (616.8 ± 48.2 s, t = -180.4, P < 0.001). Conclusions The proposed DCNN system can automatically measure mLDFA, MPTA, LDTA, JLCA, and the mechanical axis of the lower limbs, thus helping physicians manage lower limb alignment accurately and efficiently. Supplementary Information The online version contains supplementary material available at 10.1186/s12891-022-05818-4.

show abstract

Section: Discussionmentioning

confidence: 99%

Fully automated measurement on coronal alignment of lower limbs using deep convolutional neural networks on radiographic images

Meng

Wang

et al. 2022

BMC Musculoskelet Disord

View full text Add to dashboard Cite

show abstract

“…In this study, we derive a new stature estimation regression formula based on long bone measurements, which were collected from long leg radiographs of 4,200 living Europeans. Measurement was automatized by the software LAMA™ 19 , which is an algorithm able to automatically place landmarks utilizing arti cial intelligence.…”

Section: Discussionmentioning

confidence: 99%

“…For all LLRs the following linear distance measurements were computed (Fig. 5) 31 31 31 31 31 31 31 31 [19]: Leg length (measured as linear distance between top of the femoral head and midpoint of distal tibial joint line), maximum femoral length (top of the femoral head -bottom of the femoral medial condyle), and tibial length (midpoint of proximal tibial joint line -midpoint of distal tibial…”

Section: Automated Measurementsmentioning

confidence: 99%

Body height estimation from automated length measurements on standing long leg radiographs using artificial intelligence

Simon

Fischer

Rinner

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Artificial-intelligence (AI) allows large scale analyses of long-leg-radiographs (LLRs). We use this technology to derive an update for the classical regression formulae by Trotter and Gleser and Bass, which are frequently used to infer stature based on long-bone measurements. We analyzed calibrated, standing LLRs from 4,200 participants taken between 2015 and 2020. Automated landmark placement was conducted using the AI-algorithm LAMA™ and the measurements were used to determine femoral, tibial and total leg-length. Linear regression equations were subsequently derived for stature estimation. The estimated regression equations have a shallower slope and larger intercept in males and females (Femur-male: slope = 2.08, intercept = 77.49; Femur-female: slope = 1.9, intercept = 79.81) compared to the formulas previously derived by Trotter and Gleser (Femur-male: slope = 2.38, intercept = 61.41) and Bass (Femur-male: slope = 2.32, intercept = 65.53; Femur-female: slope = 2.47, intercept = 54.13). All long-bone measurements showed a high correlation (r ≥ 0.76) with stature. The linear equations we derived tended to overestimate stature in short persons and underestimate stature in tall persons. In this study, an updated regression formulae for stature estimation was established. The differences in slopes and intercepts may result from an ongoing secular increase in stature. Our study illustrates that AI-algorithms are a promising new tool enabling large scale measurements.

show abstract

“…Another application for AI in radiographs is in automated measurements, for example, leg alignment, joint orientation, leg length, or implant orientation [ 51 , 52 , 53 , 54 ]. Long-leg radiographs (LLR) are routinely acquired and manually measured to assess malalignment, which is considered a major contributing factor in OA [ 53 ].…”

Section: Methodsmentioning

confidence: 99%

“…Several studies have tested DL models on automatic measurements of the lower limb measuring HKA, AMA, and LLD. The studies found that the AI was non-inferior when compared to human readers (accuracy ranging from 89.2 to 93%) but was significantly faster [ 52 , 53 , 54 ]. However, severe deformities of the lower limb and bad image quality are common exclusion criteria for the studies mentioned above, emphasizing again that experienced clinicians are indispensable as supervisors.…”

Section: Methodsmentioning

confidence: 99%

Artificial Intelligence in Orthopedic Radiography Analysis: A Narrative Review

et al. 2022

View full text Add to dashboard Cite

Artificial intelligence (AI) in medicine is a rapidly growing field. In orthopedics, the clinical implementations of AI have not yet reached their full potential. Deep learning algorithms have shown promising results in computed radiographs for fracture detection, classification of OA, bone age, as well as automated measurements of the lower extremities. Studies investigating the performance of AI compared to trained human readers often show equal or better results, although human validation is indispensable at the current standards. The objective of this narrative review is to give an overview of AI in medicine and summarize the current applications of AI in orthopedic radiography imaging. Due to the different AI software and study design, it is difficult to find a clear structure in this field. To produce more homogeneous studies, open-source access to AI software codes and a consensus on study design should be aimed for.

show abstract

Fully automated deep learning for knee alignment assessment in lower extremity radiographs: a cross-sectional diagnostic study

Cited by 37 publications

References 26 publications

Fully automated measurement on coronal alignment of lower limbs using deep convolutional neural networks on radiographic images

Fully automated measurement on coronal alignment of lower limbs using deep convolutional neural networks on radiographic images

Body height estimation from automated length measurements on standing long leg radiographs using artificial intelligence

Artificial Intelligence in Orthopedic Radiography Analysis: A Narrative Review

Contact Info

Product

Resources

About