Introduction
Osteoporosis is a complex, multifaceted illness, classified as a civilization disease. According to data from the Polish National Health Fund for 2024, it is estimated that approximately 2.3 million people are affected by osteoporosis, of which 1.8 million are women. About 73% of this population remains undiagnosed, and the annual treatment cost for the state budget is approximately USD 371 million, with USD 308 million being| the cost associated solely with the treatment of fractures [1]. Based on a meta-analysis of 86 studies [2], it was found that 18.3% of the world’s population suffers from osteoporosis, indicating the need for improvement, enhancement, and broader diagnostic approaches in this area [2]. This disease predominantly affects the elderly, particularly postmenopausal women, accompanied by additional independent risk factors such as low body weight, low percentage of body fat, or a low body mass index (BMI). In this case, we are talking about primary osteoporosis. The secondary variant is caused by risk factors such as gastrointestinal disorders leading to malabsorption syndrome, endocrine disorders, rheumatoid arthritis, renal and liver diseases, the use of medications that disrupt bone turnover, smoking, excessive alcohol use, and insufficient calcium and vitamin D supplementation. It manifests as low-energy fractures, primarily at typical sites: the neck of the femur, the lumbar spine, and the distal third of the forearm; however, most of these fractures occur against a backdrop of low bone mass, without establishing osteoporosis. The gold standard for confirming the diagnosis and appropriate qualification remains the result of densitometric examination (dual- energy X-ray absorptiometry – DXA) at a T-score level ≤ –2.5 standard deviation (SD), performed at typical sites in the case of postmenopausal women. The T-score is the deviation from the peak bone mass in healthy individuals aged 20–29 years, expressed in standard deviations. For men aged 20–50 and premenopausal women, the Z-score is applied, which is the deviation from the norm for a given age and sex; here, a confirming result for the disease is < –2 SD [3].
The subject of this article is to describe and analyse currently available diagnostic possibilities for osteoporosis.
Material and methods
This is a literature review that was conducted on the current diagnostic possibilities for predicting osteoporosis. A literature search was performed in electronic databases including PubMed, Elsevier, and Google Scholar to identify relevant articles from 1 January 2001 to 1 October 2025. Articles were searched in databases by combining related terms such as “osteoporosis diagnostics”, “imaging diagnostics in osteoporosis” and “osteoporosis”.
Results
Dual-energy X-ray absorptiometry – the gold standard
As mentioned above, DXA remains the basis for diagnosing osteoporosis since the introduction of the method in the 1980s. The advantages of the method include relatively low costs, safety, and availability. Dual-energy X-ray absorptiometry involves examining the skeleton at typical sites (proximal femur, lumbar spine, forearm) or scanning the entire body using a low, safe dose of X-ray radiation, which constitutes 1–10% of the dose used during conventional chest X-rays. Using densitometry, one can calculate not only the bone mineral density (BMD) but also assess body composition as a three-compartment model, which includes fat mass, fat-free mass, and total body water. This model is used, among other things, for planning interventions in cases of obesity, implementing practices aimed at improving the physical condition of patients, and is also popular among athletes and fitness trainers for the correct selection of training programs [4]. Areal BMD is presented as g/cm2, or as T-score and Z-core which can confirm the diagnosis of osteoporosis, osteopenia, and indicate an increased risk of fracture. In the case of a loss of bone mass of approximately 1 standard deviation in the T-score of the lumbar spine, the risk of a fracture doubles; in the case of a similar difference in the hip, this risk increases by 2.5 times. Additionally, using DXA scans, it is possible to determine trabecular bone score (TBS), which defines the microarchitecture of bone. This parameter serves as an independent predictive tool for fractures, even when the BMD value does not directly indicate osteoporosis. The ability to assess both BMD and TBS enhances the possibility of a personalised approach to patient treatment and diagnostics [5]. However, this method has its limitations, particularly in the case of the presence of abdominal aortic atherosclerosis, intervertebral disc degeneration, and spinal deformities in relation to the examination of the lumbar vertebrae. The coexistence of these conditions may lead to misdiagnosis. This situation indicates the necessity of employing other diagnostic solutions beyond DXA [6].
X-ray – basic examination for fractures or hope for low-cost prevention?
At the end of the nineteenth century, this method revolutionized medicine and initiated imaging diagnostics. Currently, X-ray radiation is present and essential in many fields of life and science, including nuclear energy and medicine [7]. In many developing countries, it continues to be the primary method of imaging diagnostics, where there is no access to more advanced examinations, whether due to technological or economic reasons. Due to its availability and low cost, it is used as a first-line examination worldwide in the diagnosis of injuries and acute conditions of the abdominal cavity or chest. In the context of osteoporosis, X-rays are used not so much for early diagnosis of the disease but for identifying fractures that may have occurred due to the condition and for initiating detailed diagnostic and therapeutic procedures in an already developed disease. However, along with the development of technology, improvements in the quality of X-ray images, and the broader introduction of artificial intelligence (AI) into medicine, this is also changing. Through the implementation of machine learning (ML) algorithms and deep learning, we are gaining new diagnostic possibilities based on conventional X-rays [8]. These methods will be described in further detail in the following section of this article. Nevertheless, routine tests exist that can be employed as a screening method to identify patients who should be referred for further diagnostic workup, with a special role for dentists. Specifically, panoramic radiography can be used, applying Klemetti’s method to assess bone quality by evaluating the morphology of the mandibular cortex and classifying it into one of three categories: C1 (normal), C2 (mild erosion), or C3 (severe erosion). Individuals classified as C2 or C3 should be referred for DXA confirmation and further osteoporosis evaluation [9].
Computed tomography – is broad screening for osteoporosis possible?
Nowadays, computed tomography (CT) is one of the fundamental imaging diagnostic examinations that is performed very frequently. According to a study by Schoenfeld et al. [10], it was estimated that approximately 1.6–2.1 million abdominal CT scans were conducted annually in the United States from 2009 to 2011, solely for flank pain during emergency visits. The article by Young et al. [11] reported around 80 million CT scans performed annually in the United States, a significant portion of which were conducted in older adults with previously undiagnosed chronic diseases. It was noted that there was a lack of prior comprehensive utilization of data obtained from these examinations, with a focus only on the immediate cause – such as trauma. Analysing the possibilities of assessing BMD using abdominal CT scans performed for other reasons, Hounsfield units (HU) were applied to radiologically assess bone tissue density. It was determined through comparison of DXA and abdominal CT scans in the same patients that the predictive value for osteoporosis is 110 HU and below for the L1 vertebra, as noted by Pickhardt et al. [12]. Based on these assumptions, an attempt was made to determine the T-score HU for CT, which was found to be < –3.0 SD. However, it was emphasised that this should not be treated as a diagnostic criterion; rather, it is advisable to expand diagnostic measures and continue monitoring patients with such results. Utilising CT scans in this manner may prevent unnecessary fractures resulting from existing or developing osteoporosis, reduce costs for the system, and expedite diagnosis and treatment. The gold standard for diagnosis according to the WHO remains DXA [13].
Quantitative computed tomography – advanced method with insight into the past
Quantitative computed tomography (QCT) is an advanced imaging method developed for assessing volumetric bone mineral density (vBMD), as well as TBS and cortical bone mineral density. By measuring in three planes, it provides a more accurate reflection of bone strength and degenerative processes. Unlike DXA, which assesses areal BMD (in two dimensions), QCT allows for the exclusion of confounding factors such as calcifications in the aorta, spinal deformities, or visceral adipose tissue, which can interfere with the accuracy of the results. The examination involves the use of a high-resolution CT scanner, using a calibration phantom (e.g., Mindways Software phantom or the European Spine phantom) that contains calcium hydroxyapatite with a specified BMD and software that converts HU to g/cm3. Quantitative computed tomography is typically performed in relation to the spine or hip, as axial QCT, which provides volumetric BMD with trabecular and cortical compartment differentiation, improving accuracy in degenerative spine disease and enabling vertebral strength estimation, it is particularly advantageous in axial spondyloarthritis, where syndesmophytes can confound DXA. When it comes to peripheral QCT, it adds microarchitectural detail (trabecular structure, cortical thickness, porosity) at radius and tibia, yielding fracture risk information beyond hip or spine BMD and reflecting systemic bone health, relevant in the general osteoporosis risk. The accepted BMD values for osteopenia are 120–80 g/cm3 and < 80 g/cm3 for osteoporosis. Like any method, it also has its limitations. In this case, the main factor affecting errors in BMD measurement is the presence of excess bone marrow adipose tissue. One cannot overlook the higher radiation dose required compared to DXA, the relatively higher cost of the examination, as well as the equipment requirements [14, 15]. Thanks to recent developments, it is now possible to measure vBMD without the need for a phantom, utilising surrounding soft tissues as a calibrating reference, which allows for retrospective studies based on previously conducted standard CT scans, using only software [16].
Dual-energy computed tomography – dual energy, lower dose, shorter time, chemical analysis
In recent years, dual-energy computed tomography (DECT) has been gaining popularity among imaging studies. It has been successful in cardiology, rheumatology, and urology. Unfortunately, this technology is not yet widely disseminated, which hinders access for both physicians and patients. The method involves the application of two energy levels of X-ray radiation during a single examination: high and low energy. These two levels are absorbed to varying degrees by tissues due to differences in atomic numbers and densities. Both images are acquired simultaneously, which shortens the examination time and also requires a lower radiation dose. The images obtained in this manner are then overlaid, allowing for the elimination of artifacts caused by the presence of metallic foreign bodies, respiratory movements (shorter examination time compared to conventional CT), or, as mentioned in the case of QCT, excess bone marrow adipose tissue. A very important aspect is the ability to assess the chemical composition of tissues, as DECT can determine the accumulation of specific substances in the examined area, such as water, calcium, iodine, or hydroxyapatite. This technique has also found applications in the diagnosis of osteoporosis [17]. Research by Chen et al. [18] analysing their own patient group and previous studies indicates a high sensitivity and specificity of DECT in assessing BMD compared to DXA, as well as in detecting compression fractures of the vertebrae related to osteoporosis, suggesting that it could find use in screening studies in the future. However, it is noted that these are single-centre studies with a small sample size, and this issue requires larger- scale initiatives. Additionally, in the obtained image, beyond assessing vBMD, TBS, and cortical bone density, bone marrow can also be evaluated, somewhat taking on a role previously dedicated to magnetic resonance imaging (MRI). This feature improves insights into the overall structure of the bone and allows visualization of bone marrow oedema resulting from an injury that was previously undetected [19].
Magnetic resonance imaging – radiation-free alternative
Magnetic resonance imaging has emerged as a significant diagnostic alternative for assessing osteoporosis, primarily because it avoids the ionizing radiation associated with other imaging modalities such as DXA. The development of the vertebral bone quality (VBQ) score, derived from T1-weighted MRI sequences, illustrates the potential of MRI to provide valuable insights into bone health. The vertebral bone quality score quantifies the signal intensity of the vertebral body relative to cerebrospinal fluid (CSF), a method standardized across different individuals due to the consistent composition of CSF in the human population. Despite these advancements, there remains a challenge in establishing a specific VBQ threshold for definitive osteoporosis diagnosis, as existing studies report varying results based on imaging equipment and protocols. Recent research indicates that the VBQ score correlates significantly with traditional methods of assessing bone mineral density, such as DXA. Chen et al. [20] noted that VBQ achieved a diagnostic accuracy of approximately 81% for detecting osteopenia and osteoporosis when compared with DXA. In clinical practice, MRI of the lumbar region is frequently utilised to evaluate patients with low back pain or those undergoing preoperative assessments. As the role of MRI continues to expand, particularly in evaluating vertebral structure and quality, it is essential to investigate and validate the effectiveness of the VBQ score as a screening tool for osteoporosis. Preliminary findings suggest that implementing the VBQ score may improve the identification of individuals at risk, thereby facilitating timely intervention strategies. Therefore, while MRI-based VBQ represents a promising diagnostic framework, further studies will be necessary to establish diagnostic thresholds and to enhance standardization.
Quantitative ultrasound – radiation-free method for quantitative and qualitative assessment of skeletal status
Quantitative ultrasound (QUS) is a widely accessible method due to the mobility and availability of the equipment, substantially enhancing diagnostic capabilities in regions where DXA is not feasible. Additionally, this method does not expose the patient to ionizing radiation. The examination is performed on the calcaneus (heel bone), which is easily accessible, contains a high percentage of trabecular bone, and features two lateral surfaces that facilitate the propagation of ultrasound waves. The quantitative ultrasound T-score measurement can be obtained using the following parameters: speed of sound and broadband ultrasound attenuation. Speed of sound refers to the time it takes for the ultra- sound wave to traverse the length of the body part, while broadband attenuation occurs when sound waves pass through soft tissues and bone, resulting in energy absorption [21]. One of the limitations of this method is the variability in patient anatomy, especially regarding sex. Research by Yen et al. [22] shows that deviations in results were attributed to factors including the thickness of subcutaneous fat, which tends to be greater in women, affecting ultrasound wave conduction, and the surface contact area of the transducer with the skin, which is typically larger in men. Due to these and other elements that may influence QUS measurement results, it has been established as a screening tool in osteoporosis diagnostics. QUS measurements revealed the ability to express bone tissue properties in several biologic circumstances. While QUS can enhance fracture risk prediction, it cannot definitively confirm a diagnosis of osteoporosis; this requires DXA evaluation [23].
Radiofrequency echographic multi-spectrometry – fast and non-ionising method, still effective
Radiofrequency (RF) echographic multi-spectrometry (REMS) is one of the newer methods used in osteoporosis diagnosis. It consists of performing a rapid ultrasound examination using a convex 3.5 MHz transducer at typical sites, namely the hip and/or abdomen, where regions of interest are automatically localised by software, in this case the femoral neck and lumbar vertebrae. Radiofrequency echographic multi-spectrometry signals are obtained during scans. These signals are compared with reference spectral models that account for age, sex, BMI, and site, enabling classification as healthy, osteopenic, or osteoporotic, and transformation into BMD (g/cm2) and T-score/Z-score values according to the national health and nutrition examination survey database [24]. Thanks to RF signal analysis, the method, in contrast to DXA, allows exclusion of artefacts such as osteophytes, prior fractures, or metal structures. Additionally, during the examination we obtain a REMS-specific parameter, the fragility score, derived from analysis of bone microarchitecture and its integrity. Fragility score has predictive value for the occurrence of low-energy fractures within a 5-year horizon, which, together with BMD, provides a better picture than BMD alone. In a European multicentre study involving 4,307 women, REMS demonstrated a discriminative performance comparable to DXA, and even superior in individuals with artefacts. The added value of this study is certainly the safe performance in patients with contraindications to ionizing radiation tests (e.g., pregnant women) and broader accessibility in daily clinical practice due to the use of portable ultrasound devices. Unfortunately, one of the most important limitations of this method is that the examination is performed by different operators, which can lead to result variability due to lack of experience or incorrect technique. However, substantial repeatability has been demonstrated for measurements performed by the same operator (test-retest). Radiofrequency echographic multi-spectrometry can provide a valuable alternative or complement to DXA in the diagnosis and monitoring of osteoporosis, but further long-term studies and standardization of guidelines are required [25].
Deep learning – future is now
With the advancement of the internet, hardware, and software, coupled with an increasing number of users feeding databases with new information, a significant revolution in AI emerged in 2012. This marked a breakthrough in ML, particularly through deep learning (DL) approaches based on artificial neural networks. Since this pivotal moment, various fields, especially medicine, have increasingly adopted techniques to enhance efficiency and effectiveness, notably in medical imaging diagnostics, thus reducing human errors [26]. In the context of osteoporosis diagnosis, the first three publications related to DL available in the PubMed database were published in 2018. These studies addressed applying DL tools for analysing data related to diabetes-induced osteoporosis (Kruse et al. [27]), diagnosis through dental panoramic radiographs (Chu et al. [28]), and segmentation of the proximal femur from MRI images (Deniz et al. [29]). Over the subsequent years, the number of publications on this topic continued to increase year-on-year, reaching 82 by 2025, including five meta-analyses. This trend underscores the substantial significance and impact of AI in advancing scientific research. Deep learning models provide increased access to non-invasive diagnostic methods and facilitate earlier detection of osteoporosis. A study conducted by Amani et al. [30] indicates that the sensitivity, specificity, and area under the curve of DL models for diagnosing osteoporosis were 86%, 89%, and 94%, respectively, further validating these advancements.
Conclusions
The evolving landscape of osteoporosis diagnostics underscores the need for accurate and efficient methodologies to identify and manage this prevalent condition. Dual-energy X-ray absorptiometry has long been the gold standard for diagnosing osteoporosis due to its advantages such as low radiation exposure, cost- effectiveness, and capacity to provide critical information on BMD and body composition. Despite its widespread use, DXA has limitations, particularly in cases with confounding conditions like spinal deformities, indicating the necessity for complementary diagnostic methods. As medical imaging technology progresses, alternatives such as computed tomography, quantitative computed tomography, dual-energy computed tomography, MRI and REMS are increasingly explored. These modalities offer unique advantages, such as improved accuracy in assessing bone microarchitecture and BMD, while some, like MRI, provide radiation-free options that can be particularly beneficial in vulnerable patient populations. Notably, MRI’s VBQ score has shown promise in evaluating bone health and fracture risk, enhancing diagnostic capabilities by correlating well with traditional assessments like DXA. The recent introduction of AI methodologies, especially deep learning, presents exciting prospects for further revolutionising osteoporosis diagnostics. Artificial intelligence applications are positioned to not only improve the efficiency and accuracy of imaging analyses but also integrate multiple diagnostic approaches to facilitate a more comprehensive patient evaluation. The body of literature surrounding AI in medical imaging, including its application in fracture detection, highlights numerous opportunities that could reshape osteoporosis management paradigms. In conclusion, while DXA remains the cornerstone of osteoporosis diagnosis, the incorporation of advanced imaging techniques and AI into clinical practice holds significant potential to enhance diagnosis, tailor treatments, and ultimately improve patient outcomes. Continuous investment in research and technological development is essential to establish new standards and best practices in osteoporosis care.
