RESUMO
PURPOSE: An efficient physics-informed deep learning approach for extracting spinopelvic measures from X-ray images is introduced and its performance is evaluated against manual annotations. METHODS: Two datasets, comprising a total of 1470 images, were collected to evaluate the model's performance. We propose a novel method of detecting landmarks as objects, incorporating their relationships as constraints (LanDet). Using this approach, we trained our deep learning model to extract five spine and pelvis measures: Sacrum Slope (SS), Pelvic Tilt (PT), Pelvic Incidence (PI), Lumbar Lordosis (LL), and Sagittal Vertical Axis (SVA). The results were compared to manually labelled test dataset (GT) as well as measures annotated separately by three surgeons. RESULTS: The LanDet model was evaluated on the two datasets separately and on an extended dataset combining both. The final accuracy for each measure is reported in terms of Mean Absolute Error (MAE), Standard Deviation (SD), and R Pearson correlation coefficient as follows: [ S S ∘ : 3.7 ( 2.7 ) , R = 0.89 ] , [ P T ∘ : 1.3 ( 1.1 ) , R = 0.98 ] , [ P I ∘ : 4.2 ( 3.1 ) , R = 0.93 ] , [ L L ∘ : 5.1 ( 6.4 ) , R = 0.83 ] , [ S V A ( m m ) : 2.1 ( 1.9 ) , R = 0.96 ] . To assess model reliability and compare it against surgeons, the intraclass correlation coefficient (ICC) metric is used. The model demonstrated better consistency with surgeons with all values over 0.88 compared to what was previously reported in the literature. CONCLUSION: The LanDet model exhibits competitive performance compared to existing literature. The effectiveness of the physics-informed constraint method, utilized in our landmark detection as object algorithm, is highlighted. Furthermore, we addressed the limitations of heatmap-based methods for anatomical landmark detection and tackled issues related to mis-identifying of similar or adjacent landmarks instead of intended landmark using this novel approach.
Assuntos
Aprendizado Profundo , Lordose , Humanos , Reprodutibilidade dos Testes , Sacro/diagnóstico por imagem , Pelve/diagnóstico por imagem , Vértebras Lombares/cirurgiaAssuntos
Cistos Ósseos Aneurismáticos/diagnóstico , Cistos Ósseos Aneurismáticos/cirurgia , Imageamento por Ressonância Magnética , Compressão da Medula Espinal/diagnóstico , Compressão da Medula Espinal/cirurgia , Vértebras Torácicas , Seguimentos , Humanos , Laminectomia , Masculino , Pessoa de Meia-Idade , Complicações Pós-Operatórias/diagnóstico , Cirurgia Assistida por Computador , Vértebras Torácicas/cirurgia , Ultrassonografia de IntervençãoRESUMO
Calvarial diploic venous anatomy has been studied post-mortem, but few studies have addressed these venous structures in-vivo. Previous work in our laboratory has shown that intraosseous infusion through the skull diploic space near the diploic veins in animals and humans does access the superior sagittal sinus and the systemic venous system. We developed a volumetric method of imaging the diploic veins in-vivo using MRI, intravenous gadolinium, and digital subtraction to provide for three-dimensional depiction and exact localization of these veins. We hypothesized that this technique would allow for an assessment of the probability of existence, distribution, and concentration of diploic veins in the skull. We scanned 31 neurosurgical patients, and were able to create 3D diploic venous maps in 74% of them. These maps were processed using Adobe Photoshop CS2. Mathworks MatLab 6.5, once customized, counted the number of pixels occupied by the diploic veins in the processed image. The probability of veins was highest in the occipital regions (100%). The inferior occipital (4.1%) and posterior parietal (4.1%) regions had the highest concentrations of diploic veins. Digital subtraction venography using a volumetric MRI sequence can demonstrate the diploic veins in-vivo. The inferior occipital region may be the best area for an intraosseous infusion device because it has the greatest likelihood of containing a vein and also has the highest concentration of veins.
Assuntos
Angiografia Digital/métodos , Angiografia por Ressonância Magnética/métodos , Crânio/irrigação sanguínea , Adolescente , Adulto , Idoso , Animais , Meios de Contraste , Feminino , Gadolínio DTPA , Humanos , Processamento de Imagem Assistida por Computador , Injeções Intravenosas , Masculino , Pessoa de Meia-Idade , Veias/anatomia & histologia , Adulto JovemRESUMO
Hemicraniectomy and opening underlying dura mater permits the expansion of infarcted, swollen brain outwards, reversing dangerous intracranial pressure elevations and the risk of fatal transtentorial temporal lobe or diencephalic herniation. Recently published randomized controlled trials have proven this procedure a powerful life-saving measure in the setting of malignant middle cerebral artery infarction and allayed concerns that a reduction in mortality is accompanied by an unacceptable increase in patients suffering severe neurological impairments. Appropriate patients are relatively young, in the first five decades of life, suffering infarction of a majority of the middle cerebral artery (MCA) territory in either hemisphere, and decompression should be performed prior to progression to coma or two dilated, fixed pupils. Lethargy combined with midline shift and uncal herniation on neuroimaging is an appropriate trigger to consider and discuss surgical intervention. Families and, when possible, patients themselves, should be informed of the certainty of at least moderate to mild permanent deficits, and the possibility of worse. To be successful decompression must be extensive, targeting a bone flap measuring 14 cm from front to back, and extending 1 to 2 cm lateral to the midline sagittal suture to the floor of the middle cranial fossa at the level of the coronal suture. An augmentation duraplasty is mandatory.