Artificial intelligence-based orthopaedic perpetual design

Abstract

Background and aims

Integrating Artificial Intelligence (AI) methodologies in orthopaedic surgeries is becoming increasingly important as it optimises implant designs and treatment procedures. This research article introduces an innovative approach using an AI-driven algorithm, focusing on the humerus bone anatomy. The primary focus of this work is to determine implant dimensions tailored to individual patients.

Methodology

We have utilised Python's DICOM library, which extracts rich information from medical images obtained through CT and MRI scans. The algorithm generates precise three-dimensional reconstructions of the bone, enabling a comprehensive understanding of its morphology.

Results

Using algorithms that reconstructed 3D bone models to propose optimal implant geometries that adhere to patients' unique anatomical intricacies and cater to their functional requirements. Integrating AI techniques promotes enhanced implant designs that facilitate enhanced integration with the host bone, promoting improved patient outcomes.

Conclusion

A notable breakthrough in this research is the ability of the algorithm to predict implant physical dimensions based on CT and MRI data. The algorithm can infer implant specifications that align with patient-specific bone characteristics by training the AI model on a diverse dataset. This approach could revolutionise orthopaedic surgery, reducing patient waiting times and the duration of medical interventions.

1. Introduction

1.1. Background and gaps in the literature

Orthopaedics specialises in diagnosing, treating, preventing, and rehabilitating musculoskeletal disorders and injuries. This surgical branch has witnessed significant advancements and potential future scope, addressing various musculoskeletal diseases and injuries; interventions, such as joint replacements/arthroplasty, spine surgeries, fracture fixation, and deformity corrective procedures, are being formed faster healing to minimise long-term complications. Reconstructive orthopaedic surgery is a highly specialised field that requires a profound understanding of medical imaging, anatomy, and orthopaedics. Such software development typically involves close collaboration with medical professionals, engineering, and software professionals to ensure its accuracy and suitability for clinical use. Natural language processing (NLP) is a widely utilised AI instrument within the healthcare sector, which extracts variables and performs classifications.¹

AI-driven medical imaging analysis like MRI and X-rays has empowered orthopaedic experts to precisely identify and diagnose issues such as fractures, joint irregularities, and tumours. Rapid analysis of extensive imaging data by AI algorithms supports medical professionals in making highly accurate diagnoses and devising targeted treatment strategies.

The spotlight turns toward (AI) in orthopaedics, where research primarily centres on spinal, knee, and hip-related matters. However, using AI in therapeutic applications and sub-specialities still exhibits constraints. In orthopaedics, conventional joint replacement surgery is based on some specific standards that are not personalised. However, the patterns of fractures from patient to patient may pose constraints for implant selection. These constraints are strict regulations, difficulty accessing high-quality data, regulatory approval, compatibility with existing systems, and resistance to change. There is an imminent requirement for establishing standardised reporting for AI research and initiating prospective trials to address this gap.² Advancements in surgical techniques and technology, such as robotic-assisted surgeries (RAS) and arthroscopy, are expected to reduce invasiveness, promote faster recovery, and minimise scarring. However, many Orthopaedic surgeons are aware of AI use but are still determining its safety, and hence, research is needed to assess its usefulness and safety.³

Biomedical agents, personalised medicine AI, implant technology, remote monitoring, tissue engineering, nanotechnology, preventive strategies, and global access to care are also expected to revolutionise orthopaedic treatments. AI integration in orthopaedic surgery focuses on ethical and legal considerations, patient autonomy, privacy, and workforce shifts.⁴ Combining AI algorithms, data analytics, and innovative technology can improve patient outcomes, increase efficiency, and transform care delivery, which is the importance of AI in delivering value-based healthcare.⁵ Additive manufacturing (AM) is revolutionising the medical industry with customised orthotics, 3D printing, and biocompatibility.⁶

Newer orthopaedic implants pose challenges due to their biocompatibility, durability, time-consuming manufacturing, and sensory feedback. Emerging technologies like smart prosthetics, 3D printing and AI can enhance functionality. Interdisciplinary collaboration, increased research investment, component standardisation, and improved access are crucial for success.⁷ Prosthetic joint infection (PJI) after total hip arthroplasty (THA) may require debridement, implant retention, and antibiotic therapy, with atypical microorganisms challenging.⁸ A study analysed 145 orthopaedic surgeons' intentions to integrate new medical technologies, revealing that perceived advantages, risks, quality, experience, and receptivity to digital tools influence their use.⁹ In modern Orthopaedic treatments, options include various innovative surgical techniques and technological innovations involving multidisciplinary approaches.¹⁰ Industry 5.0 is particularly appealing in highlighting its transformative technologies that enable personalised treatments by seamlessly integrating machinery and human expertise.¹¹ Industry 5.0 helps develop individualised products for diagnosing, treating, and managing various orthopaedic conditions. AI has significantly impacted Orthopaedics with technological advancements and access to handheld devices. Industry 4.0, through smart manufacturing, can meet mass production needs, but it may only partially meet personalised implantation. Through the integration of automation and the enhancement of labour efficiency, Industry 5.0 has facilitated the creation of patient-tailored tools, instruments, and implants aimed at enhancing clinical outcomes, functional improvements, and Patient-Related Outcome Measures (PROMs).12, 13, 14 Computer technology and implant design advancements have led to developing smart instruments and intelligent implants in trauma and orthopaedics, improving patient-related functional outcomes. Sensor technology uses embedded devices to detect physical, chemical, and biological signals, offering diagnostic capabilities and therapeutic benefits. These implants have applications in total knee arthroplasty, hip arthroplasty, spine surgery, fracture healing, early detection of infection, and implant loosening. Smart sensor implant technology objectively assesses ligament and soft tissue balancing, maintains sagittal and coronal alignment, and achieves desired kinematic targets. Post-implantation data can monitor implant performance and patient clinical recovery during rehabilitation.¹⁵ MRI and PET scans are used to evaluate articular cartilage integrity, with PET-MRI combining them for detailed joint imaging, potentially aiding in understanding OA pathophysiology.¹⁶ The Internet of Medical Things (IoMT) combines medical devices and applications with healthcare information technology systems, offering improved care and satisfaction for orthopaedic patients during the COVID-19 pandemic. It enables data sharing, patient tracking, and remote-location healthcare, transforming healthcare facilities and improving patient satisfaction.^17,18 Healthcare organisations require AI-driven decision-making technologies to manage COVID-19 and prevent its spread. AI mimics human intelligence, aiding in real-time suggestions and vaccine development. It aids in screening, analysing, and tracking patients, including confirmed, recovered, and death cases.¹⁹ 3D printing aids in complex trauma management by accurately reducing implant placement, reducing surgical time, and improving outcomes. Although initial learning curves exist, practice and experience make these techniques easier.20, 21, 22, 23 Recent research and publications show a surge in interest in 3D printing in orthopaedic surgery despite its primitive stage due to insufficient knowledge, high costs, and learning curve, suggesting its potential for future orthopaedics and trauma cases.^24,25 This review explores the need for technological advancements and the critical challenges in finding innovative solutions.

1.2. Aims

We aim to highlight the areas where orthopaedic practices can be reshaped by investigating manufacturing techniques, streamlined processes, and regulatory enhancements and the integration of AI-based technology with the following principal objectives.

• •To enhance the orthopaedic prosthesis selection process by developing an algorithm to improve decision-making efficiency and accuracy.
• •To develop a technique to regenerate the numerous bone structures using CT and MRI data.
• •To create a 3D reconstruction of the humerus bone and develop a method for accurate implantation.

2. Methodology

Data collection involves gathering a diverse dataset of medical images and preprocessing them to enhance quality and standardised formats. Image segmentation uses AI-based techniques to identify target structures and generate accurate 3D models of the patient's anatomy.^26,27

Extracts of relevant features from the models and clinical data are achieved by feature extraction, while data labelling and annotation serve as the foundation for training and validating the AI algorithm (Fig. 1). Machine learning model development uses convolutional neural networks (CNN) or recurrent neural networks (RNN) to develop predictive models, and personalised design optimisation involves inputting a patient's specific anatomical data into the trained model.²⁸

Fig. 1.

Artificial Intelligence (AI)-based design Algorithm.

Validation and testing are conducted using real-world patient cases and clinical scenarios, evaluating factors like implant stability, alignment accuracy, and surgical outcomes. Continuous learning and improvement mechanisms are implemented, with feedback from orthopaedic surgeons and refinement based on new data to enhance accuracy and effectiveness.

Integration into the surgical workflow involves user-friendly interfaces, seamless integration into existing software, and adherence to medical device regulations and ethical guidelines. Python is a versatile programming language widely employed in various fields, including medical imaging. Regarding medical imaging, the Digital Imaging and Communications in Medicine (DICOM) standard is crucial for storing and transmitting medical images.²⁹

Python's libraries, Pydicom, MONAI, Vedo, and Nibabel, enable seamless interaction with medical image files, allowing for efficient manipulation and analysis.³⁰ Python's integration with DICOM plays a pivotal role in image reconstruction. It facilitates data extraction from DICOM files, reconstructing three-dimensional images from a series of DICOM slices. This capability is pivotal in diverse medical applications, from CT scans to magnetic resonance imaging (MRI), enhancing diagnosis and treatment planning.

Fig. 2 shows the trans-axial view, sagittal view and coronal view representing partial bone structure, and the skeleton part generated using the algorithm is shown in Fig. 3.

Fig. 2.

View generated from DICOM data (A-Trans-axial view; B-Sagittal view; C-Coronal view).

Fig. 3.

Skeleton generated using Artificial Intelligence (AI) algorithm in Python.

After generating the skeleton bone structure, for example, humerus bone may be extracted from the generated surface, as shown in Fig. 4. Fig. 4(a) represents the point cloud data of humerus bone extracted from the skeleton structure, and it is further processed for surface modelling. The rough surface model thus generated from point cloud data is shown in Fig. 4(b), and after further processing for a smooth surface, the bone model is seen in Fig. 4(c). The quality of the surface depends upon the density of point cloud data.

Fig. 4.

Humerus bone structure from Artificial Intelligence (AI) - generated surface.

This approach optimises patient-specific solutions, offering multiple options for joint support and expediting treatment selection based on real-time data, ultimately enhancing healthcare outcomes.

3. Results and discussion

The design and implementation of orthopaedic implants are time-consuming processes, alignment and external support design problems, instant design and specialised design availability. Time-taking processes can delay timely treatment, especially in urgent cases like fractures and malignancies. Customised implant designs may be necessary due to patient anatomy or complex injuries, and the lack of these designs can compromise treatment outcomes. Proper alignment is crucial for implant success, and issues like joint instability, limited range of motion (ROM), and premature implant failure can arise. Designing implants that integrate seamlessly with external supports can be challenging, and the immediate availability of appropriate designs is essential for avoiding complications in urgent cases. Specialised implant designs tailored to unique patient needs may be available slowly.³¹

3.1. AI-based 3D reconstruction

AI-based methods excel in rapidly and precisely reconstructing bones using patient-specific DICOM or CT scan images, streamlining the process compared to manual programming. The resultant 3D models play a crucial role across various applications, from crafting 3D-printed prototypes for fracture treatment to supporting diverse analyses, research and clinical decision-making. These breakthroughs significantly contribute to refining bone implant design and optimisation, ensuring a better fit and enhanced functionality for patients.

These innovations empower surgeons to conduct surgeries more precisely, improving patient outcomes. Surgeons can utilise precise 3D models to meticulously plan and execute procedures, thereby increasing success rates and diminishing associated surgical risks. The convergence of programming and AI-driven techniques in 3D reconstruction catalyses reshaping healthcare practices, effectively bridging the gap between technical expertise and medical advancements. The future of this research imagines an instantaneous implant recommendation system bolstered by cloud-based data infrastructure, which would harness the power of AI and cloud computing to process and analyze medical imaging data rapidly, expediting treatment decisions.³²

The progressive transformation of implantation techniques can be discerned as a reflection of the burgeoning medical cognisance and the concurrent advancement of technological capabilities. The 21st century has witnessed a paradigm shift in the assimilation of digital paradigms, advanced imaging modalities, and the application of personalised medical treatments. This transformative era engendered the integration of computer-assisted surgery (CAS), patient-specific implants (PSI), additive manufacturing (AM) through 3D printing, and the deployment of AI-informed surgical planning. Cumulatively, these advancements are instrumental in the recalibration of modern orthopaedic practices, thereby fostering a dynamic milieu that is attuned to the amelioration of patient outcomes and an overall improvement in the quality of life.³³

3.2. Personalised implant design

Personalised solutions ensure that implant designs and surgical guides are optimised to fit each patient's distinct anatomy, reducing design time and workload during surgery.³⁴ In making evidence-based decisions, the algorithm utilises data-driven insights to guide surgical choices, enhancing precision and minimising complications. Innovative design solutions employ advanced computational techniques, and the algorithm continually learns from real-world cases to improve its recommendations.³⁵ Advancements in cloud, internet, AI imaging, and 5G technologies have made digital healthcare relevant in various clinical indications and applications.³⁶ A successful implementation necessitates close collaboration between orthopaedic surgeons, medical device engineers, data scientists, and regulatory experts, guaranteeing the algorithm's safety, accuracy, and clinical effectiveness.³⁷ The cost and the delay time for implant manufacturing in the case of personalised implants can be further calculated with the process and the machinery availability with its capacity for precision and degree of flexibility. We understand that entering the Orthopaedic ‘market’ with AI-based perpetual designs involves considerations in cost, technological advancement, and practicality. The initial development cost of advanced AI technologies may be high, however, as technology advances, costs could decrease, making it more feasible for the users.

3.3. Radiological aspects

Imaging studies are pivotal in Orthopaedic care. Manifestations. Baseline radiological assessments rely on plain radiographs, whereas Computed Tomography (CT) and Magnetic Resonance Imaging (MRI) scans are employed to assess deeper clinical aspects such as bony fusion, hardware integrity, and the presence of implant loosening.³⁸ The imaging is immensely useful for spinal problems in preoperative diagnosis and intra-operative and postoperative imaging.^39,40 Thirty patients who had metallic implants were assessed using CT scans along with multiplanar reconstruction (MPR), and the results indicated that the application of MPR substantially decreased metal-related artefacts in transaxial images.⁴¹ The Syngo Explorer is a new clinical X-ray computed tomography software application for routine and scientific work. It allows users to reconstruct, process, and view CT images independently, using the Syngo platform for patient data management and visualisation.⁴² The advanced single-slice rebinning (ASSR) algorithm is proposed for medical spiral cone-beam CT systems with large detector rows. It uses virtual reconstruction planes tilted to fit 180° spiral segments, achieving high image quality, low patient dose, and low reconstruction times. The algorithm's computational complexity is comparable to standard single-slice CT, allowing for available 2D back projection hardware.⁴³

3.4. Comparison of programming-based techniques vs. AI-based methods

Programming-based approaches necessitate substantial proficiency and knowledge in 3D reconstruction techniques, making them less accessible to individuals without specialised training. Traditional programming methods can be intricate and time-consuming, potentially leading to errors in the reconstruction process due to the complexity of the algorithms. Programming-based approaches often require more automation and adaptability than AI-based methods, potentially requiring manual adjustments and interventions during reconstruction. Table 1 compares Programming-based techniques and AI-based methods.⁴⁴

Table 1.

Salient features of comparison of Programming-based techniques vs. AI-based methods.

	PRE-REQUISITES	PROS	CONS
Programming-based techniques	Necessitate substantial proficiency and knowledge in 3D reconstruction techniques.	Precise Control	Less accessible to most
		Clear Debugging	Time-consuming and complex
		Domain Knowledge	Likely errors possible
Artificial Intelligence-based methods	Can efficiently reconstruct bones using patient-specific DICOM or CT scan images	-The resulting model serves as a foundational element for many applications, from creating 3D-printed prototypes for fracture mending to facilitating diverse analyses and elevating the design and optimisation of bone implants.	Data Dependency,bvbv
		-AI can automate and create a similar design of bone that is difficult to program explicitly, leading to increased efficiency.	Ethical issues,
		-AI can identify complex patterns of bone from CT or MRI scan images and correlations in large datasets that might be too challenging for humans to discern.	Interpretability problems

These breakthroughs permeate various medical contexts, ultimately offering support to surgeons in executing procedures with heightened precision and seamless ease. As the landscape of medical advancements progresses, the convergence of programming and AI-driven techniques in 3D reconstruction emerges as a driving force, reshaping the equilibrium between technical expertise and the evolution of medical practices toward a more accessible and adaptive horizon.

Creating AI-based orthopaedic perpetual designs involve integrating AI algorithms to continuously optimize orthopaedic implants, prosthetics, or devices for enhanced performance and patient outcomes. The AI system can analyze patient data, biomechanics, and material science to iteratively refine designs, ensuring adaptability to evolving medical knowledge and individual patient needs. Entering the market with AI-based perpetual designs in Orthopaedics necessitates managing costs, leveraging technological advancements, and ensuring practicality for seamless integration into the healthcare landscape.

4. Limitations and future scope

This paper discusses AI-based orthopaedic implant designs, which highlights the potential benefits of this approach. We acknoweldge its limitations such as machine variability, accuracy, and challenges in scalability and standardisation. This paper also emphasises the need for rigorous validation through ANSYS or other software simulations to accurately represent real-world performance. The need for clear development criteria for personalised implants, acknowledging clinical implementation challenges, and addressing 3D reconstruction, manufacturing time and cost, including the need for ethical and legal considerations, patient consent, data privacy, and liability, to ensure responsible implementation of personalised implant technologies. Technologically, AI-driven perpetual designs in Orthopaedics can enhance diagnostic accuracy, treatment planning, and personalised care. The balance lies in ensuring that the technology aligns with regulatory standards and provides tangible benefits over traditional approaches. Implementing AI in Orthopaedics requires collaboration with healthcare professionals, integration into existing systems. Striking a balance between innovation and practicality is also crucial for successful market adoption.

5. Conclusion

The programming-based techniques demand an exhaustive comprehension and skillset for reconstructing from DICOM or other medical images. They also come with the accessibility challenge, often confined to individuals with specialised training. Their complexity and time-consuming nature can render them prone to errors in the reconstruction process due to intricate algorithms. Absence of automation and adaptability intrinsic to AI-based methods can necessitate manual interventions and adjustments throughout the reconstruction process. Conversely, AI-based methods can efficiently reconstruct bones using patient-specific DICOM or CT scan images. The resulting model serves as a foundational element for many applications, from creating 3D-printed prototypes for fracture mending to facilitating diverse analyses and elevating the design and optimisation of bone implants.

Conflict of interest

Nothing to disclose.

Funding

None.

Ethical approval

Not required.

Use of AI tools

None.

Authors’ contribution

MNA: Manuscript Writing, Programming and processing, Editing and Final approval, AH: Manuscript Writing, Literature Search, Conceptulaization, Editing and Final approval, MJ: Manuscript Writing, Organisation and Data collection, Editing and Final approval, SK: Manuscript Writing, Data Collection, Methodology, Editing and Final approval, AV: Manuscript Writing, Conceptualization, Technical modifications Editing and Final approval, RV: Manuscript Writing, Conceptualization, Literature search, Technical modifications, Editing and Final approval.

Declaration of competing interest

None.