Anatomy is a unique subject where one must learn by doing. This is achieved through a variety of methods like dissection, use of prosections, imaging, playful learning (crocheting structures, drawing, and painting organs), through functional movement like yoga or pilates, use of 3D plastic models or by immersing oneself in virtual or augmented realities. Experiential learning, as described by Kolb, that is "learning by doing," underpins all practical aspects of learning this important but fact dense subject. Reflective observation, abstract conceptualization and active experimentation follow on from the concrete experience, first usually experienced either in the dissecting room or through living anatomy. We map out how the educator can best support the learner to move from one stage to the next to gain the most out of their learning experiences in anatomy. Though anatomy is rich and diverse in the ways it can be taught, we focus on dissection and encountering anatomy through imaging to illustrate how experiential learning theory permeates everyday teaching and learning practices. Many opinions have been offered on dissection as a learning method, which we believe is foundational for healthcare students learning the subject, as well as the importance of offering radiology within the anatomy curriculum and we hope to illustrate how the two are mutually dependent and not exclusive.
This report traces the authors' long journey toward a deeper understanding of the anatomical basis of rheumatologic physical examination. Fifty years ago, one of the authors was struck by the prevalence of subcutaneous and deep bursitis and regional musculoskeletal pain syndromes at his new workplace, prompting studies of bursal swelling, lubrication, and the dynamics of distended superficial and deep bursae. Subsequently, a deficiency in anatomical knowledge was noted across multiple workshops at the American College of Rheumatology and elsewhere. Later, in a different setting, musculoskeletal anatomy workshops were held, based on cross-examination between rheumatology learners and instructors. Palpation was used in the upper extremity, and the spine and lower extremity were demonstrated through motion and self-palpation. Pre-workshop tests consistently revealed poor recall of musculoskeletal anatomy among trainees and practitioners, including orthopedic trainees. All questions were rated highly important in an international Delphi survey of anatomical items relevant to rheumatology. Subsequently, a self-examination method was developed to allow learners to practice, in privacy, the anatomical knowledge traditionally acquired. More recently, an international, unfunded, and enthusiastic study group comprising anatomists, rheumatologists, and ultrasonographers made independent observations of defined neck and upper extremity structures, including the elusive lower belly of omohyoid, the dorsal forearm muscle-tendon intersection, the concurrent contraction of anconeus and pronator teres, and dynamic aspects of the extensor apparatus of the fingers, the natatory ligament, the web spaces, and the intertendinous connections. In these studies, physical examination and ultrasound complemented each other. Thus, physical examination, rather than being replaced, should be strengthened. Key Points • We describe a sustained effort spanning more than 50 years to reinstate MSK anatomy as a basic competence for MSK conditions and general rheumatologic examination. • The initial studies focused on subcutaneous and deep bursae and on seminars regarding musculoskeletal pain syndromes and, much later, the anatomical basis of rheumatologic examination. • This was followed by a one-on-one assessment of rheumatology fellows' and practitioners' anatomical knowledge and by participatory MSK anatomy seminars held across the Americas with support from ILAR. • Subsequently, a method of self-examination was developed as an accessible way to reinforce anatomical learning. • Currently, independent physical examination and ultrasonography, along with dissection performed by anatomists, have resulted in a series of innovative and fruitful proof-of-concept studies.
暂无摘要（点击查看详情）
Ligaments and mesenteries divide the peritoneal cavity into interconnected compartments. In addition to the underlying disease, the anatomical conditions of the abdominal cavity and the intraperitoneal distribution of fluid have a decisive influence on the course and spread of the illness. Knowledge of these factors provides insight into clinically relevant disease processes and thus enables systematic diagnosis. Ligamente und Mesenterien unterteilen die Peritonealhöhle in untereinander kommunizierende Kompartimente. Neben der zugrundeliegenden Erkrankung haben die anatomischen Verhältnisse der Bauchhöhle und die Flüssigkeitsverteilung intraperitoneal einen entscheidenden Einfluss auf den Krankheitsverlauf und die Krankheitsausdehnung. Die Kenntnis von Anatomie und Flüssigkeitszirkulation erlaubt einen Einblick in klinisch relevante Krankheitsprozesse und ermöglicht damit eine systematische Befundung.
To review the evolution of sleep medicine and surgical management of obstructive sleep apnea (OSA), focusing on anatomy-based and individualized treatment strategies. This narrative review was developed by an expert panel of otolaryngologists and sleep medicine specialists. Landmark studies, consensus statements, and clinically relevant investigations were analyzed. Surgical approaches were classified according to anatomical target and functional mechanism, with particular attention to drug-induced sleep endoscopy (DISE), patient phenotyping, and multimodal treatment strategies. Sleep surgery has progressed from highly invasive procedures to minimally invasive, function-preserving, and multilevel approaches. Advances in understanding upper airway anatomy and collapse mechanisms have enabled targeted interventions, including barbed pharyngoplasty, transoral robotic surgery, maxillomandibular advancement, and hypoglossal nerve stimulation. DISE has improved identification of obstruction patterns and individualized surgical planning. Current OSA management increasingly combines surgical, dental, behavioral, and medical therapies within personalized care pathways. Modern sleep surgery has evolved toward precision-based, multidisciplinary management. Individualized anatomical and functional assessment is central to treatment selection, while emerging diagnostic and therapeutic innovations may further improve outcomes.
In the late 1960s, Professor M. Gazi Yaşargil began to systematically incorporate microsurgery into neurosurgical practice, grounded in anatomical expertise and intensive, systematic laboratory training followed by clinical application. Using the operative microscope, he introduced microsurgical techniques and redefined surgical anatomy that had long existed but remained largely unseen. By exploiting natural cisternal pathways to achieve pure lesionectomy, he reintroduced operability for complex pathologies and established new standards in modern neurosurgery. Yaşargil's early contributions to microvascular training and neurosurgical practice exemplify enduring principles from which neurosurgical professionals at all stages can continue to draw guidance. Our exposure to his work-through lectures, publications, and personal communications-profoundly shaped our own approach to neurosurgery and ultimately informed the philosophy underlying our global neurosurgery efforts, embodied in the Madison Microneurosurgery Initiative. This manuscript reviews the historical foundations of microneurosurgery through Yaşargil's early career, his formative laboratory work in Burlington, Vermont, and the systematic clinical implementation of microsurgery in Zurich, distilling key lessons derived from these pioneering experiences. A companion manuscript (Part II) describes how these principles were translated into structured curricula and implemented through contemporary microsurgical training and our global neurosurgery efforts.
In the Forensic Collection of the Institute of Forensic Medicine in Belgrade there is an autopsy specimen of a man who underwent total reconstruction of the penis he had previously lost due to war injury, collected in the year 1929 by Professor Milovan Milovanović (1884-1948). The museum exhibit No. 453, labeled as Penis Arteficialis, is a specimen of in toto dissected male pubic region, including the grossly altered male external genitalia - the mechanical injury that completely destroyed the penis was partially overcome by a procedure of total phalloplasty: abdominal skin flap was used to create a bulky phalloid structure, with autologous rib cartilage fragment incorporated in its proximal part to provide sufficient rigidity for sexual intercourse. The procedure left an acquired hypospadia, since its inherent limitation was the inability to reconstruct the urethra through the artificial penis. The unaltered autopsy specimen was analyzed using multi-detector computed tomography and magnetic resonance imaging to reveal the complete anatomy of the pioneering total phalloplasty, while preserving the museum specimen. So, we demonstrated the essential elements of what was a successful reconstructive procedure of the time: the artificial penis with cartilage proximally embedded into the root portion of the erectile tissue, enabling stimulation, arousal, and rigidity, and the unobstructed urethra that opened on the skin below the neophallus. Milovanović sought to preserve this achievement in reconstructive surgery, and today, with imaging, we can directly observe the pioneering surgical approach to total penile reconstruction, appreciating its achievements and limitations in terms of morphology.
Autosomal dominant polycystic kidney disease (ADPKD), primarily driven by PKD1 or PKD2 mutations, is a prevalent hereditary nephropathy for which large-animal models capturing human renal anatomy and disease tempo remain urgently needed. This study aimed to establish a site-specific human PKD2 (hPKD2) transgenic porcine model as a versatile platform capable of supporting two temporally distinct applications: longitudinal evaluation of whether sustained PKD2 overexpression is sufficient to induce renal pathology, and-should overexpression prove phenotypically silent-future crossbreeding-based functional rescue of PKD2-knockout lines. A CRISPR/Cas9-mediated, homology-recombination-independent strategy was employed to target full-length hPKD2 cDNA into the porcine pH11 safe harbor locus. Somatic cell nuclear transfer yielded five F0 transgenic founders, and natural mating of two founders with wild-type sows produced four F1 transgenic offspring, confirming stable germline transmission. Quantitative real-time PCR and whole-genome sequencing validated single-copy, site-specific transgene integration at the designated locus. Robust hPKD2 mRNA and FLAG-tagged polycystin-2 (PC-2) expression were detected in renal and other tissues across both generations. During the initial 12-month monitoring period, serum blood urea nitrogen and creatinine levels remained within normal ranges and no gross histological abnormalities were evident. However, given that these biomarkers are insensitive to early-stage renal impairment, extended observation with more comprehensive phenotyping is required before definitive conclusions regarding renal function can be drawn. A transgenic porcine model with stable, single-copy hPKD2 integration at the pH11 safe harbor locus was successfully generated and shown to permit germline transmission. This platform provides a foundation for long-term investigation of PKD2 overexpression pathophysiology and, alternatively, for functional complementation of PKD2-deficient models, thereby advancing both mechanistic and translational ADPKD research.
Naviculocuneiform (NC) arthrodesis is indicated for multiple pathologies, including degenerative and traumatic arthritis, progressive collapsing foot deformity, and medial column instability. Despite its broad application, no consensus exists regarding the optimal fixation construct. Shape-memory dynamic compression staples offer potential advantages in midfoot fusion compared to static fixation. To evaluate the feasibility of using 4-legged dynamic compression staples for NC arthrodesis and to characterize NC joint anatomy to guide implant selection and orientation. Level IV cadaveric study. Twenty fresh-frozen cadaveric lower limbs (10 pairs) were dissected to expose the NC joint. Measurements of the medial and intermediate cuneiform and navicular facets were recorded. Four-legged staples (26 and 30 mm) were inserted, and adjacent joints were examined for proximal or distal penetrance. Statistical analysis was performed using Mann-Whitney U tests with significance set at P ≤ .05. Staple fixation across the medial cuneiform-navicular articulation was safe in all 40 applications without talonavicular or tarsometatarsal breakthrough. In contrast, the intermediate segment demonstrated a significantly shorter length (35.83 ± 3.24 mm vs 46.40 ± 4.15 mm; P = .0002) and higher rates of talonavicular breakthrough at 20% for 26 mm staples and 60% for 30 mm staples. Longer intermediate segments correlated with reduced breakthrough (P = .0190). To our knowledge, this is the first study to quantify NC articulation lengths. Four-legged staples up to 30 mm in length were shown to be feasible for use in both dorsal-to-plantar or medial-to-lateral orientations at the medial cuneiform articulation. Due to breakthrough with the 4-legged staples, 2-legged staples (16-20 mm bridge) and staple legs ≤20 mm are advised for the intermediate segment to minimize iatrogenic injury risk.Level of Evidence: Level IV Cadaveric Study.
To describe a multidisciplinary protocol for the resection and reconstruction of complex lateral skull base pathology, based on over two decades of experience in a high-volume tertiary referral center in the United Kingdom. Retrospective descriptive analysis of institutional practice. A high-volume tertiary skull base unit in the United Kingdom. This study presents a summary of clinical experience in managing lateral skull base pathologies-both benign and malignant-over a 20-year period. It outlines a multidisciplinary approach involving ENT, neurosurgery, plastic surgery, anesthesia, radiology, oncology, and specialist nursing teams. The institutional protocol for resection and reconstruction is described, including surgical planning, intraoperative decision-making, perioperative care, and long-term rehabilitation strategies. The integrated multidisciplinary protocol enabled safe and effective management of a broad range of lateral skull base lesions. The team achieved high rates of complete resection and functional reconstruction, with acceptable levels of morbidity. Complex anatomy, tumor extent, and reconstructive requirements were successfully navigated through coordinated team-based planning. The approach has been refined through iterative experience and now serves as a reference pathway for similarly structured centers. Effective management of lateral skull base pathology requires a highly coordinated multidisciplinary strategy. The described protocol, developed and implemented over two decades, offers a structured and reproducible approach that may serve as a valuable reference for other high-volume centers involved in skull base surgery.
Current preclinical models for cervical dysplasia rely on animal systems that poorly mimic human anatomy. To address this and align with emerging initiatives to reduce animal experimentation, we developed a human-sized, 3D in vitro model of high-grade cervical dysplasia. The construct integrates normal human fibroblasts, keratinocytes, and SiHa cancer cells within a gelatin methacrylate (GelMA) hydrogel, engineered with a cervical "os" and asymmetric lesion to replicate native human cervical architecture. We utilized this platform to evaluate the efficacy and safety of ethyl cellulose-ethanol (EC-ethanol), a novel ablation therapy designed to mitigate the off-target leakage associated with traditional ethanol injections. While control injections of pure ethanol resulted in widespread, non-specific necrosis (approximately 99% cell death), EC-ethanol successfully formed a localized gel depot. High-resolution mapping of the ablation zone demonstrated that EC-ethanol significantly concentrated cytotoxicity within the dysplastic lesion while preserving surrounding healthy tissue. Margin analysis revealed a sharp therapeutic gradient, with cell death normalizing to background levels within approximately 2 mm of the injection site. These findings validate the translational potential of EC-ethanol as a spatially precise intervention and demonstrate the utility of macro-scale 3D models as powerful, ethical alternatives to animal testing for optimizing local drug delivery.
Gamma Knife radiosurgery (GKRS) is an established treatment for vestibular schwannoma, but contemporary cohort-level data integrating long-term tumor control with clinical symptom trajectories and hearing-related correlates remain valuable. We performed a retrospective cohort study of 286 patients with vestibular schwannoma (VS) treated with GKRS (2014-2024). The primary endpoint was local control. Secondary outcomes included overall survival and changes in hearing loss, imbalance, vertigo, dizziness, and tinnitus. Kaplan-Meier analysis was used to estimate local control and overall survival. Paired symptom changes were assessed using McNemar testing. The Median follow-up was 52.3 months (IQR 27.4-80.3). Progressive radiographic enlargement occurred in 5 patients, yielding a crude local control rate of 98.3%. The actuarial local control was 98.5% at 5 years and 96.5% at 10 years. Imbalance, vertigo, dizziness, and tinnitus improved significantly after GKRS, whereas hearing loss remained stable overall (59.4% vs. 57.3%, p = 0.441). Mean symptom burden declined from 1.74 to 1.20 symptoms per patient (p < 0.001). Post-radiosurgery hearing loss was more frequent in Koos IV tumors and increased stepwise across quartiles of cochlear mean dose: 38.7% for ≤ 2.9 Gy, 53.3% for > 2.9-3.6 Gy, 66.2% for > 3.6-4.5 Gy, and 73.8% for > 4.5 Gy (p = 0.00012). Baseline hearing loss remained the strongest predictor in multivariable modeling. At a median of four years GKRS provided tumor control in > 95% of VS patients and was associated with reduced overall symptom burden and improvement in vestibular symptoms. Hearing outcomes were related to baseline auditory status, Koos grade, and cochlear dosimetry. Not applicable.
Stroke is a severe cerebrovascular disease characterized by narrow time windows and complications. This study aimed to identify novel drug targets and repurposed drugs for stroke. This study used expression quantitative trait loci data from druggable genes in brain and blood as instrumental variables. Mendelian randomization, colocalization, and phenome-wide Mendelian randomization were applied to evaluate causal relationships and potential side effects, with stroke and ischemic stroke as primary outcomes. Preclinical validation used oxygen-glucose deprivation/reperfusion and middle cerebral artery occlusion/reperfusion models. Pharmacological and behavioral assessments evaluated the therapeutic potential of candidate targets and drugs. Additionally, proteomic sequencing was performed following GGCX (γ-glutamyl carboxylase) overexpression to explore its biological functions. Elevated GGCX expression in brain and blood was potentially causally associated with reduced risk of stroke and ischemic stroke, supported by colocalization evidence, although potential cardiovascular risks could not be excluded. Drug repositioning identified ifenprodil as a candidate agent that reduced infarction volume, improved motor and cognitive functions, and reversed GGCX downregulation in mice. Ifenprodil treatment and GGCX overexpression alleviated oxygen-glucose deprivation/reperfusion-induced injury and upregulated GGCX expression. Mechanistically, GGCX conferred neuroprotection by regulating protein homeostasis, suppressing inflammation, promoting metabolic recovery, and modulating nuclear transcriptional regulation. This study established a potential causal link between GGCX and stroke risk, particularly ischemic stroke. GGCX represents a promising therapeutic target for ischemic stroke. Targeted GGCX expression upregulation and drug repurposing, particularly ifenprodil, may offer novel therapeutic avenues. Further validation is warranted to assess clinical efficacy and safety.
The currently known favorable properties of soybean have become the target of fraud in meat production. The use of plant-based raw materials in the adulteration of meat products is increasingly recognized by consumers, manufacturers, and researchers. However, the toxicity of soybeans has not been addressed. This article aimed to focus on the following: detection, quantification of soybean adulteration in meat products, analyzing how other adulterants could impact cholesterol levels in the products, and discussing how soybean adulteration could cause conditions such as allergies and toxicity. Utilizing immunohistochemical techniques, 450 different meat product samples were analyzed. Highlighting the confirmed soybean adulteration by analyzing cholesterol levels using HPLC. Results showed that, across all tested cases, soybean toxic doses ranged from 62% to 85%. Meanwhile, there were no indications on their respective nutritional facts labels. Surprisingly, the cholesterol levels were lower than those reported by previous investigators.
Zika virus (ZIKV) infections have been linked to severe neurological disorders, including microcephaly and Guillain-Barré syndrome in humans as well as mouse models of ZIKV infection. Despite the association, the mechanisms underlying ZIKV-induced neuropathology remain incompletely understood. We have recently shown that antigen independent CD8+ T cells mediate neurological disease in ZIKV-infected mice independent of the amount of infectious virus in the CNS. To further investigate the role of brain viral load and lymphocytes in ZIKV infection we studied the viral kinetics, pathology, and immune responses of ZIKV-infected NOD-Rag1-/-Il2rg-/- mice, which are deficient in lymphoid cells. Despite prolonged high viral titers in the brain, NOD-Rag1-/-IL2rg-/- mice did not develop neurological symptoms following ZIKV infection, contrasting with the infection outcomes of Ifnar1-/- mice which exhibit paralysis despite lower viral load. Notably, we observed significant differences in brain myeloid cells in the presence or absence of lymphoid cells. While Ifnar1-/- mice showed robust infiltration of CD45hiCD11b+ cells in the brain, lymphocyte-deficient NOD-Rag1-/-IL2rg-/- mice exhibited reduced recruitment and activation of these cells. Additionally, we found that CD45hiCD11b+ cells displayed a more inflammatory phenotype in Ifnar1-/- mice compared to NOD-Rag1-/-IL2rg-/- mice. Our study highlights the complex interplay between the immune system and viral infection in ZIKV-induced neuropathology and underscores the importance of considering immune responses in the development of therapeutic interventions for ZIKV.
Neuronopathic Gaucher disease (nGD) is a lysosomal storage disorder caused by GBA1 mutations, leading to defective acid β-glucosidase (GCase) and accumulation of glycosphingolipid substrates, causing inflammation and neurodegeneration. Patients with nGD manifest severe neurological symptoms, but current animal models fail to fully recapitulate the human condition, posing a major barrier to the development of effective therapies targeting the brain. To bridge this gap, we have developed midbrain-like organoids (MLOs) from human induced pluripotent stem cells of nGD patients with GBA1L444P/P415R and GBA1L444P/RecNcil mutations to model nGD brain pathogenesis. These nGD MLOs exhibited GCase deficiency, resulting in diminished enzymatic function, accumulation of lipid substrates, widespread transcriptomic changes, and impaired dopaminergic neuron differentiation, mirroring nGD pathology. GBA1 mutation correction mediated by CRISPR/Cas9 restored GCase activity, normalized lipid substrate levels, and rescued dopaminergic neuron function, confirming the causal role of GBA1 mutations during early brain development. Using this novel platform, we further evaluated therapeutic strategies, including SapC-DOPS nanovesicles delivering GCase, AAV9-GBA1 gene therapy, and substrate reduction therapy with GZ452, a glucosylceramide synthase inhibitor currently under clinical investigation. These treatments either restored GCase activity, reduced lipid substrate accumulation, improved autophagic and lysosomal abnormalities, or ameliorated dysregulated genes involved in neural development. These patient-specific, 3D neural models offer a transformative, physiologically relevant platform for unraveling disease mechanisms and accelerating the discovery of therapies for patients with nGD.
ObjectiveSkin cancer diagnosis via automated image analysis remains a challenging task due to poor image contrast, visual similarity among lesion classes, and class imbalance in available datasets. To address these issues, this study proposes a novel Tri Model Dermatology Cancer Neural Network (TriDermCancerNet) for classifying skin cancer from dermoscopic images.MethodsTwo publicly available datasets are used in this work: the International Skin Imaging Collaboration 2018 and 2019 datasets, which contain multiple classes of cancer types. In the proposed model framework, a contrast enhancement technique was applied to improve image quality, followed by data augmentation to balance the datasets. The proposed TriDermCancerNet architecture is designed based on the key challenges of this work, including dataset variability, interclass similarity, and model explainability. The proposed architecture integrates three modules: a 105-layer Inception module, a 186-layer Inverted Bottleneck Residual module, and the Dense-177 module. Each module was integrated into the proposed network in parallel rather than in series. Thereafter, each module was trained, and features were extracted and fused using a depth concatenation approach. During training, several important hyperparameters were selected using Bayesian optimization, and the final model was used for classification in the testing phase.ResultsThe fused TriDermCancerNet achieved 98.6% accuracy, 98.6% sensitivity, 98.6% F1-score, and an area under the curve of 1.0 on the International Skin Imaging Collaboration 2018 dataset and 99.7% accuracy, 99.6% sensitivity, 99.6% F1-score, and an area under the curve of 1.0 on the International Skin Imaging Collaboration 2019 dataset. Statistical significance testing confirmed that the fusion model outperforms each branch (p < 0.05).ConclusionThe proposed hybrid TriDermCancerNet approach enhances the precision and robustness of skin cancer classification frameworks, providing clinicians with a valuable diagnostic aid for early detection.
Parkinson's disease (PD) is a common neurodegenerative disease. Our previous single-cell sequencing results suggested that Transcription Factor E3 (TFE3) was only differentially expressed in microglia in the MPTP mouse model. However, the functional role of microglial TFE3 in PD pathogenesis remains unclear. First, motor function was assessed in MPTP mice following TFE3 overexpression in substantia nigra microglia. Second, RNA-seq was used to identify the function of TFE3 in microglia, and the molecular mechanism was verified both in vivo and in vitro. Finally, we investigated whether TFE3 nuclear translocation affects its transcriptional activity and subsequently influences microglia in vitro. Overexpression of TFE3 in substantia nigra microglia could alleviate PD-related phenotypes. RNA-seq revealed that TFE3 regulates microglial phagocytosis and inflammation. Mechanistically, TFE3 affects these functions by regulating the expression of Mer receptor tyrosine kinase (Mertk) and lysosomal-associated membrane protein 1 (Lamp1). Finally, Rapamycin could activate the nuclear translocation of TFE3 and enhance phagocytosis and alleviate inflammation of microglia. Our findings demonstrate that Rapamycin activates TFE3, which in turn upregulates the expression of Mertk and Lamp1 in the substantia nigra. This TFE3-mediated pathway plays a critical role in regulating microglial phagocytosis and inflammation in the PD model.
暂无摘要（点击查看详情）
To investigate the feasibility of using color fundus photographs (CFPs) combined with artificial intelligence (AI) for detecting diabetic macular ischemia (DMI), a condition characterized by retinal capillary loss in the macula that leads to vision impairment in diabetic patients. Despite the widespread use of CFPs and AI in diagnosing other eye diseases like diabetic retinopathy (DR), their application in DMI detection remains unexplored because of skepticism regarding its viability. A graph neural network-based multispectral-view learning (GNN-MSVL) model was developed to detect DMI from CFPs. The model uses computational multispectral imaging to reconstruct 24-wavelength pseudo-multispectral fundus images from CFPs, enhancing sensitivity to subtle reflectance changes caused by ischemic tissue. ResNeXt101 served as the backbone for multi-view feature extraction, whereas a customized GNN with jumper connections improved cross-spectral relationship learning. The study included 1078 macula-centered CFPs from 1078 eyes of 592 diabetic patients, with 530 images from 300 patients confirmed as DMI cases. The GNN-MSVL model achieved an accuracy of 84.7% and an area under the receiver operating characteristic curve of 0.900 (95% confidence interval, 0.852-0.937) at the eye level, significantly outperforming both baseline CFP-trained models and human experts (P < 0.01). AI-based analysis of CFPs shows promising potential for DMI detection, offering a feasible, early, and cost-effective screening method. This approach could address the current gap in DMI diagnosis and improve clinical outcomes for diabetic patients. This study will establish a feasible, CFP-based screening method for DMI.