1987 Fiscal Year Final Research Report Summary
A study of the optimum image processing and display methods of digital chest X-ray image.
| Project/Area Number |
61440048
|
|---|
| Research Category |
Grant-in-Aid for General Scientific Research (A)
|
| Allocation Type | Single-year Grants |
|---|
| Research Field |
Radiation science
|
|---|
| Research Institution | Keio University |
|---|
Principal Investigator |
HASHIMOTO Shozo School of Medicine, Keio University Professor, 医学部, 教授 (40050348)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
OGAWA Koichi School of Medicine, Keio University Assistant, 医学部, 助手 (00158817)
ANDO Yutaka School of Medicine, Keio University Lecturer, 医学部, 講師 (20118904)
YUTA Shinichi Institute of Information Science and Electronics University of Tsukuba Associate, 電子情報工学系, 助教授 (00092502)
NAKAJIMA Masato Faculty of Science and Technology, Keio University Associate Professor, 理工学部, 助教授 (20051766)
|
|---|
| Project Period (FY) |
1986 – 1987
|
| Keywords | Digital chest X-ray image / Medical image processing / Image data compression / Edge enhancement / Contrast enhancement / Look-up table control / 特徴抽出 / 診断能 |
|---|
| Research Abstract |
We examined a laser image scanner for an X-ray film digitalization unit. As the result, the density of film digitized by the laser image scanner was 2/3 of that digitized by a micro-densitometor, and the spatial resolution of laser image scanner was 2-31p/mm. As concerns an optimum display, we studied the control of window level and window width of a gray scale. For the display of a digital chest X-ray image, the mean density and the standard deviation was useful for the setting of the window level and width. We could get the optimum display with the window level (mean-SD/2) and window width (4SD). An inverse display of gray scale was also useful for the display of the lung field. Moreover we controlled the look-up table of the CRT. The result showed improvements of the contrast of a diagnostic image. As concerns the image processing, we researched the contrast enhancement method, i.e., the adaptive histogram equalization method, and the edge enhancement method, i.e., the unsharp mask method. The output image after these image processings, had different kinds of information that was useful for diagnosis compared with the ordinary X-ray film image. It implies the possibility of a new diagnostic radiology. We studied also the evaluation method of a diagnostic ability. An artificial coin lesion(ACL) was proposed for the purpose and it was confirmed the validity by superimposed on an actual digital chest X-ray image. We researched the image data compression method and the data compression using the block quantized discreet cosine transform(DCT) was examined. From the results, the threshold sampling was superior to another sampling method, and the block size of 16x16, 32x32 were suitable for quantization. The evaluation of decompress image was performed using the score method and the reciever operating characteristics(ROC) analysis. Consequently, the data compression ratio of 1/5 or 1/10 were acceptable for the diagnosis.
|
