Search

The deep feedback Group Method of Data Handling (GMDH)-type neural network is applied to the medical image diagnosis of lung cancer. The deep feedback GMDH-type neural network can identified very complex nonlinear systems using heuristic self-organization method which is a type of evolutionary computation. The deep neural network architectures are organized so as to minimize the prediction error criterion defined as Akaike’s Information Criterion (AIC) or Prediction Sum of Squares (PSS). In this algorithm, the principal component-regression analysis is used for the learning calculation of the neural network. It is shown that the deep feedback GMDH-type neural network algorithm is useful for the medical image diagnosis of lung cancer because deep neural network architectures are automatically organized using only input and output data.

The deep Group Method of Data Handling (GMDH)-type neural network is applied to the medical image analysis of brain X-ray CT image. In this algorithm, the deep neural network architectures which have many hidden layers and fit the complexity of the nonlinear systems, are automatically organized using the heuristic self-organization method so as to minimize the prediction error criterion defined as Akaike’s Information Criterion (AIC) or Prediction Sum of Squares (PSS). The learning algorithm is the principal component-regression analysis and the accurate and stable predicted values are obtained. The recognition results show that the deep GMDH-type neural network algorithm is useful for the medical image analysis of brain X-ray CT images.

The objective of this study was to evaluate the usefulness of transcatheter arterial embolization prior to surgical excision of musculoskeletal tumors. We reviewed the records of nine patients (3 females and 6 males) who received arterial embolization prior to excision of musculoskeletal tumors in our hospital from December 2009 to April 2010. We evaluated tumor region, size, histopathology, feeding artery, embolic material, and blood loss during surgery. We compared the actual amount of intraoperative bleeding with arterial embolization to estimated amounts of bleeding without embolization predicted by three orthopedic surgeons. Arterial embolization was performed on the same day or within 5 days before surgery. Operations were performed as planned in all patients without serious complications. The amount of intraoperative bleeding was 35-4200 mL and there was significantly less bleeding with arterial embolization compared with the estimated amounts (p<0.01). Our results show that arterial embolization prior to resection of hypervascular musculoskeletal tumors reduces the amount of bleeding during surgery and contributes to patient safety.

Purpose : The purpose of this study was to evaluate the usefulness of thick slabminimum intensity projection (MinIP) as a follow-up method in patients with pulmonaryemphysema. This method was used to determine the presence or absence of changes overtime in the lung field based on multi-detector-row CT (MDCT) data. Methods : Among patientsdiagnosed with pulmonary emphysema who underwent 16-MDCT (slice thickness,1 mm) twice at an interval of 6 months or more, 12 patients without changes in the lungfield and 14 with clear changes in the lung field were selected as subjects. An image interpretationexperiment was performed by five image interpreters. Pulmonary emphysemawas followed up using two types of thick slab MinIP (thick slab MinIP 1 and 2) andmulti-planar reformation (MPR), and the results of image interpretation were evaluatedby receiver operating characteristic (ROC) analysis. In addition, the time required forimage interpretation was compared among the three follow-up methods. Results : Thearea under the ROC curve (Az) was 0.794 for thick slab MinIP 1, 0.778 for the thick slabMinIP 2, and 0.759 for MPR, showing no significant differences among the three methods.Individual differences in each item were significantly more marked for MPR thanfor thick slab MinIP. The time required for image interpretation was around 18 secondsfor thick slab MinIP 1, 11 seconds for thick slab MinIP 2, and approximately 127 secondsfor MPR, showing significant differences among the three methods. Conclusion : Therewere no significant differences in the results of image interpretation regarding the presenceor absence of changes in the lung fields between thick slab MinIP and MPR. However,thick slab MinIP showed a shorter image interpretation time and smaller individualdifferences in the results among image interpreters than MPR, suggesting the usefulnessof this method for determining the presence or absence of changes with time in the lungfields of patients with pulmonary e

We studied the lumbar spines of 117 adults (39 women and 78 men) with spondylolysis unrelated to low back pain using multidetector computed tomography (CT). Of the 117 subjects with spondylolysis, including five with multiple-level spondylolysis, there were 124 vertebrae with spondylolysis. In adult lumbar spines with unilateral spondylolysis, there was no significant difference between the incidence of spondylolisthesis in female and male subjects. However, in those with bilateral spondylolysis, there was a significantly higher incidence of spondylolisthesis in female subjects (90.9%) than in males (66.2%). Furthermore, females with bilateral spondylolysis had significant more slippage than males. Lumbar index and lumbar lordosis were not significantly different between male and female subjects, and did not significantly correlate with slippage. In conclusion, to treat acute spondylolysis in adolescents, it is important to obtain bony union at least unilaterally, especially in female subjects, to prevent further slippage.