Proximal femur segmentation and quantification in dual-energy subtraction tomosynthesis: A novel approach to fracture risk assessment

Abstract

Background

Osteoporosis is a major public health concern, especially among older adults, due to its association with an increased risk of fractures, particularly in the proximal femur. These fractures severely impact mobility and quality of life, leading to significant economic and health burdens.

Objective

This study aims to enhance bone density assessment in the proximal femur by addressing the limitations of conventional dual-energy X-ray absorptiometry through the integration of tomosynthesis with dual-energy applications and advanced segmentation models.

Methods and Materials

The imaging capability of a radiography/fluoroscopy system with dual-energy subtraction was evaluated. Two phantoms were included in this study: a tomosynthesis phantom (PH-56) was used to measure the quality of the tomosynthesis images, and a torso phantom (PH-4) was used to obtain proximal femur images. Quantification of bone images was achieved by optimizing the energy subtraction (ene-sub) and scale factors to isolate bone pixel values while nullifying soft tissue pixel values. Both the faster region-based convolutional neural network (Faster R-CNN) and U-Net were used to segment the proximal femoral region. The performance of these models was then evaluated using the intersection-over-union (IoU) metric with a torso phantom to ensure controlled conditions.

Results

The optimal ene-sub-factor ranged between 1.19 and 1.20, and a scale factor of around 0.1 was found to be suitable for detailed bone image observation. Regarding segmentation performance, a VGG19-based Faster R-CNN model achieved the highest mean IoU, outperforming the U-Net model (0.865 vs. 0.515, respectively).

Conclusions

These findings suggest that the integration of tomosynthesis with dual-energy applications significantly enhances the accuracy of bone density measurements in the proximal femur, and that the Faster R-CNN model provides superior segmentation performance, thereby offering a promising tool for bone density and osteoporosis management. Future research should focus on refining these models and validating their clinical applicability to improve patient outcomes.

Keywords

osteoporosis proximal femur tomosynthesis dual-energy subtraction faster R-CNN U-Net

Get full access to this article

View all access options for this article.

References

Michael Lewiecki

Watts

McClung

, et al. Official positions of the international society for clinical densitometry. J Clin Endocrinol Metab 2004; 89: 3651–3655.

Wasnich

Ross

Davis

, et al. A comparison of single and multi-site BMC measurements for assessment of spine fracture probability. J Nucl Med 1989; 30: 1166–1171.

Ross

Davis

Epstein

, et al. Pre-existing fractures and bone mass predict vertebral fracture incidence in women. Ann Internal Med 1991; 114: 919–923.

Johnell

Gullberg

Kanis

, et al. Risk factors for hip fracture in European women: the MEDOS study, mediterranean osteoporosis study. J Bone Miner Res 1995; 10: 1802–1815. doi:https://doi.org/10.1002/jbmr.5650101125

Ohama

Sanada

Nakagawa

. Photodensitometry: microdensitometry (MD): digital image processing method (DIP). Nihon Rinsho 1994; 52: 2317–2322.

Langton

Palmer

Porter

. The measurement of broadband ultrasonic attenuation in cancellous bone. Eng Med 1984; 13: 89–91.

Black

Cummings

Genant

, et al. Axial and appendicular bone density predict fractures in older women. J Bone Miner Res 1992; 7: 633–638.

Genant

Engelke

Fuerst

, et al. Noninvasive assessment of bone mineral and structure: state of the art. J Bone Miner Res 1996; 11: 707–730.

Iki

Kajita

Mitamura

, et al. Precision of quantitative ultrasound measurement of the heel bone and effects of ambient temperature on the parameters. Osteoporos Int 1999; 10: 462–467.

10.

Chappard

Berger

Roux

, et al. Ultrasound measurement on the calcaneus: influence of immersion time and rotation of the foot. Osteoporos Int 1999; 9: 318–326.

11.

Dobbins

Godfrey

. Digital x-ray tomosynthesis: current state of the art and clinical potential. Phys Med Biol 2003; 48: R65–R106.

12.

Shiomi

. The principle and clinical application of tomosynthesis. Jpn J Imaging Inform Sci Med 2007; 24: 22–27.

13.

Ichikawa

. CT super basic, Ohmsha, 2015.

14.

Barrett

Keat

. Artifacts in CT: recognition and avoidance. Radiographics 2004; 24: 1679–1691.

15.

Machida

Yuhara

Tamura

, et al. Whole-Body clinical applications of digital tomosynthesis. RadioGraphics 2016; 36: 735–750.

16.

Fogelman

Blake

. Different approaches to bone densitometry. J Nucl Med 2000; 41: 2015–2025.

17.

Blake

Fogelman

. Technical principles of dual energy x-ray absorptiometry. Semin Nucl Med 1997; 27: 210–228.

18.

Takeuchi

. SONIALVISION G4 high-performance R/F system and its tomosynthesis, SLOT advance, and bone density measurement applications for high quality diagnosis. Medical Now 2018; 83: 40–42.

19.

Astaraki

Severgnini

Milan

, et al. Evaluation of localized region-based segmentation algorithms for ct-based delineation of organs at risk in radiotherapy. Phys Imaging Radiat Oncol 2018; 5: 52–57.

20.

Gupta

Anand

. A hybrid edge-based segmentation approach for ultrasound medical images. Biomed Signal Process Control 2017; 31: 116–126.

21.

Chowdhary

Mittal

Kumaresan

, et al. An efficient segmentation and classification system in medical images using intuitionist possibilistic fuzzy C-mean clustering and fuzzy SVM algorithm. Sensors 2020; 20: 3903. doi:https://doi.org/10.3390/s20143903

22.

Jena

Jain

Nayak

, et al. Analysis of depth variation of U-NET architecture for brain tumor segmentation. Multim Tools Appl 2023; 82: 10723–10743.

23.

Kayalibay

Jensen

van der Smagt

. CNN-based segmentation of medical imaging data. arXiv preprint 2017. doi:https://doi.org/10.48550/arXiv.1701.03056

24.

Ren

Girshick

, et al. Faster R-CNN: towards real-time object detection with region proposal networks. IEEE Trans Pattern Anal Mach Intell 2017; 39: 1137–1149.

25.

Gkioxari

Dollár

, et al. Mask R-CNN. IEEE Trans Pattern Anal Mach Intell 2020; 42: 386–397.

26.

Zhiqiang

Jun

. A review of object detection based on convolutional neural network. 36th Chinese Control Conference, 2017, pp. 11104–11109. doi:https://doi.org/10.23919/ChiCC.2017.8029130

27.

Ronneberger

Fischer

Brox

. U-Net: convolutional networks for biomedical image segmentation, international conference on medical image computing and computer-assisted intervention. Lecture Notes Comp Sci 2015; 9350: 234–241.

28.

Siddique

Paheding

Elkin

, et al. U-net and its variants for medical image segmentation: a review of theory and applications. IEEE Access 2021; 9: 82031–82057.

29.

Zhou

Siddiquee

MMR

Tajbakhsh

, et al. UNet++: A nested U-Net architecture for medical image segmentation. Deep learning in medical image analysis and multimodal learning for clinical decision support, 2018, pp. 3–11.

30.

Cao

Wang

Chen

, et al. Swin-Unet: Unet-like pure transformer for medical image segmentation. Computer vision ECCV 2022 workshops, 2022, pp. 205–2018. doi:https://doi.org/10.1007/978-3-031-25066-8_9

31.

Yıldız Potter

Yeritsyan

Mahar

, et al. Proximal femur fracture detection on plain radiography via feature pyramid networks. Sci Rep 2024; 14: 12046.

32.

Bjornsson

Baker

Fleps

, et al. Fast and robust femur segmentation from computed tomography images for patient-specific hip fracture risk screening. Comp Methods Biomech Biomed Eng Imaging Visualizat 2023; 11: 253–265.

33.

Deng

Wang

Zhao

, et al. A deep learning-based approach to automatic proximal femur segmentation in quantitative CT images. Med Biol Eng Comput 2022; 60: 1417–1429.

34.

C-T

Matsushima

Huang

Y-H

, et al. Harnessing yolo algorithms for efficient breast cancer detection in mammography. J Med Imaging Radiat Sci 2023; 54: S7.

35.

Matsushima

Chen

T-B

Hsu

S-Y

, et al. Investigation of a recognition system for general X-ray images using CNN and faster R-CNN. J Signal Process 2022; 26: 159–169.

36.

Matsushima

Kimura

Tai-Been Chen

, et al. Extraction and quantification of fracture prediction images of the proximal femur using CNN. NCSP'24 Proceedings, 2024, pp. 344–347.

37.

Bengio

. Practical recommendations for gradient-based training of deep architectures. Neural networks, 2012, pp. 437–478. Springer.

38.

Iguchi

. Studies of bone mineral density in aged patients with femoral neck fracture. Cent Jpn J Orthop Surg Traumatol 1999; 42: 507–514.