• Radiation Oncology Journal
• /
• 제35권2호
• /
• pp.180-184
• /
• 2017

Patients with locally advanced lung cancer and very limited pulmonary function (forced expiratory volume in 1 second $[FEV1]{\leq}1L$) have dismal prognosis and undergo palliative treatment or best supportive care. We describe two cases of locally advanced node-positive non-small cell lung cancer (NSCLC) patients with very limited lung function treated with induction chemotherapy and moderate hypofractionated image-guided radiotherapy (Hypo-IGRT). Hypo-IGRT was delivered to a total dose of 45 Gy to the primary tumor and involved lymph nodes. Planning was based on positron emission tomography-computed tomography (PET/CT) and four-dimensional computed tomography (4D-CT). Internal target volume (ITV) was defined as the overlap of gross tumor volume delineated on 10 phases of 4D-CT. ITV to planning target volume margin was 5 mm in all directions. Both patients showed good clinical and radiological response. No relevant toxicity was documented. Hypo-IGRT is feasible treatment option in locally advanced node-positive NSCLC patients with very limited lung function ($FEV1{\leq}1L$).

• 전자공학회논문지B
• /
• 제31B권8호
• /
• pp.160-171
• /
• 1994

The wide dynamic range and severely attenuated contrast in mediastinal area appearing in typical chest radiographs have often caused difficulties in effective visualization and diagnosis of lung diseases. This paper proposes a new adaptive image enhancement technique which potentially solves this problem and there by improves observer performance through image processing. In the proposed method image processing is applied to the chest radiograph with different processing parameters for the lung field and mediastinum adaptively since there are much differences in anatomical and imaging properties between these two regions. To achieve this the chest radiograph is divided into the lung and mediastinum by gray level thresholding using the cumulative histogram and the dynamic range compression and local contrast enhancement are carried out selectively in the mediastinal region. Thereafter a gray scale transformation is performed considering the JND(just noticeable difference) characteristic for effective image displa. The processed images showed apparenty improved contrast in mediastinum and maintained moderate brightness in the lung field. No artifact could be observed. In the visibility evaluation experiment with 5 radiologists the processed images with better visibility was observed for the 5 important anatomical structures in the thorax.

• 방송공학회논문지
• /
• 제18권5호
• /
• pp.702-712
• /
• 2013

본 논문은 흉부 CT 영상에서 폐 흉벽에 결절 및 폐혈관이 붙어 있는 경우에도 폐 정보의 손실 없이 폐와 기관지를 분리할 수 있는 알고리즘을 제안 하였다. 마스크 영상의 활용은 폐 및 기관지 분할에서 시간 단축 및 성능을 향상 시킬 수 있었다. 또한 폐 흉벽과 밝기값이 같은 결절을 찾아 제거 하는 방법은 좌 우측폐의 외곽 영상을 2진 영상으로 변환하고, 형태학적 정보를 활용함으로써 가능 하였다. 마지막으로 제거된 부분의 외곽선 연결은 거리가 고려된 최적 화소 추가와 3차 Spline 보간법을 적용하였다. Matlab 시뮬레이션 결과 제안된 알고리즘은 기존 문제점이 보완됨을 확인 할 수 있었다.

• 전자공학회논문지
• /
• 제51권2호
• /
• pp.173-181
• /
• 2014

본 논문은 흉부 CT에서 폐, 기관지 및 폐혈관을 자동으로 분할 할 수 있는 알고리즘을 제안 하였다. 제안된 방법은 3단계로 진행된다. 첫째는 최적 임계값과 3차원 레이블링을 통한 영역성장법으로 폐 및 기관지를 분할한다. 둘째는 기관지의 형태학적 정보를 활용하여 기관지의 첫 번째 분기점(용골)까지는 차감연산으로, 이후부터는 가변적 임계값 기법을 적용하여 기관지를 분할한다. 셋째는 폐에 대한 복원 과정으로 좌/우측 폐를 분리하고, 개선된 롤링 볼 알고리즘을 적용하여 폐 외곽의 이상 유무를 확인하며, 이상이 발견되면 그 부분을 제거하고, 2차 다항식 형태로 폐 외곽을 연결시킴으로서 정상적인 폐로 복원한다. 마지막으로 폐혈관은 임계 값 기법의 3 차원 레이블링과 3 차원 영역성장법을 적용하여 분할하였다. 시뮬레이션 결과 폐 주변조직의 손실 없이 정확하게 분할됨을 확인 할 수 있었다.

• Advances in nano research
• /
• 제12권2호
• /
• pp.185-195
• /
• 2022

Deep learning is another field of artificial intelligence (AI) utilized for computer aided diagnosis (CAD) and image processing in scientific research. Considering numerous mechanical repetitive tasks, reading image slices need time and improper with geographical limits, so the counting of image information is hard due to its strong subjectivity that raise the error ratio in misdiagnosis. Regarding the highest mortality rate of Lung cancer, there is a need for biopsy for determining its class for additional treatment. Deep learning has recently given strong tools in diagnose of lung cancer and making therapeutic regimen. However, identifying the pathological lung cancer's class by CT images in beginning phase because of the absence of powerful AI models and public training data set is difficult. Convolutional Neural Network (CNN) was proposed with its essential function in recognizing the pathological CT images. 472 patients subjected to staging FDG-PET/CT were selected in 2 months prior to surgery or biopsy. CNN was developed and showed the accuracy of 87%, 69%, and 69% in training, validation, and test sets, respectively, for T1-T2 and T3-T4 lung cancer classification. Subsequently, CNN (or deep learning) could improve the CT images' data set, indicating that the application of classifiers is adequate to accomplish better exactness in distinguishing pathological CT images that performs better than few deep learning models, such as ResNet-34, Alex Net, and Dense Net with or without Soft max weights.

• 대한방사선기술학회지:방사선기술과학
• /
• 제35권3호
• /
• pp.249-254
• /
• 2012

본 논문은 유방암 환자의 접선 조사 시, 치료 전 수행하는 electronic portal image와 치료 계획용 2-D reconstruction image를 비교하는 방법으로 자세의 정확성을 평가하고자 한다. 방사선 치료 중 접선조사(tangential breast treatment)만을 받는 22명의 유방암 환자를 대상으로 자세 정렬의 정확성을 확인 하였다. electronic portal image와 치료 계획용 2-D reconstruction image의 해부학적 기준 매개 변수를 비교하여 그 오차 도를 평가 하였다. 접선조사 환자의 44매 2-D reconstruction image와 110매의 EPID image 상의 비교 기준 매개 변수는 치료 조사면 중심부(field center)의 폐 길이, CLD(central lung distance), 치료 조사면 중심부의 연부조직 길이, CSTD(central soft tissue distance), 상부 총 폐 길이, ALD (above lung distance), 하부 총 폐 길이, BLD(below lung distance)이며, 내측 접선조사면(medial tangential field)에서 각 매개 변수의 오차 평균값은 1.0, -6.4, -2.1, 2.0, 각각의 표준편차(${\sigma}$)는 1.5, 2.3, 4.1, 1.1 이다. 외측 접선조사면(lateral tangential field)의 각 매개 변수 오차 평균값은 -1.5, -4.3, -2.7, -1.3 이며, 각각의 표준편차(${\sigma}$)는 3.3, 2.1, 2.9, 그리고 2.5로 나타났다. 접선조사 치료를 받는 유방암 환자의 EPID image 상에서 CLD, CSTD, ALD 그리고 BLD의 인식은 매우 용이하며 이를 근거로 자세 정렬 오차를 판단하는 것이 시간과 숙련도의 단축을 이끌어 낼 수 있다고 사료된다.

• 한국의학물리학회:학술대회논문집
• /
• 한국의학물리학회 2002년도 Proceedings
• /
• pp.442-445
• /
• 2002

We present an algorithm for automatic anatomically adaptive image enhancement of digital chest radiographs. Chest images were exposed using digital radiography system with a 0.143 mm pixel pitch, l4-bit gray levels, and 3121 ${\times}$ 3121 matrix size. A chest radiograph was automatically divided into two classes (lung field and mediastinum) by using a maximum likelihood method. Each pixel in an image was processed using fuzzy domain transformation and enhancement of both the dynamic range and local gray level variations. The lung fields were enhanced appropriately to visualize effectively vascular tissue, the bronchus, and lung tissue, etc as well as pneumothorax and other lung diseases at the same time with the desired mediastinum enhancement. A prototype implementation of the algorithm is undergoing trials in the clinical routine of radiology department of major Korean hospital.

• Asian Pacific Journal of Cancer Prevention
• /
• 제16권14호
• /
• pp.5613-5618
• /
• 2015

Clinicians should looking for techniques that helps to early diagnosis of cancer, because early cancer detection is critical to increase survival and cost effectiveness of treatment, and as a result decrease mortality rate. Medical images are the most important tools to provide assistance. However, medical images have some limitations for optimal detection of some neoplasias, originating either from the imaging techniques themselves, or from human visual or intellectual capacity. Image processing techniques are allowing earlier detection of abnormalities and treatment monitoring. Because the time is a very important factor in cancer treatment, especially in cancers such as the lung and breast, imaging techniques are used to accelerate diagnosis more than with other cancers. In this paper, we outline experience in use of image processing techniques for lung and breast cancer diagnosis. Looking at the experience gained will help specialists to choose the appropriate technique for optimization of diagnosis through medical imaging.

• 한국정보통신학회논문지
• /
• 제13권10호
• /
• pp.2163-2170
• /
• 2009

흉부 컴퓨터단층영상에서의 얻어진 폐 영상은 볼륨과 형태 등의 정량적인 정보들로서 진단과 수술 계획 등에 있어서 필연적 정보를 제공한다. 일반적인 영상분할은 이미지를 구성 요소영역이나 목적물에 따라 나누는 방법이다. 그러나 재분할을 하는 단계에서 최종영상은 에너지 최소화를 해결하는 정도에 의존하며, 분할은 응용대상의 관심 영역에서 객체나 물체의 경계에서 정지하게 된다. 가변형 능동모델은 컴퓨터 비젼, 영상처리 분야에서 광범위하게 사용되고 있다. 또한 영역 분할은 현재까지 많은 연구가 되고 있으며, Xu에 의해서 GVF라는 새로운 형태의 외부힘이 제안되고 있다. 본 논문에서 제안하는 알고리듬은 흉부 컴퓨터단층영상에서 실질을 자동 분할하기 위해서 에너지 최소화 방법을 사용하고, 영역분할을 위해 개선된 가변형 능동모델을 제안한다. 알고리듬은 정확한 영역분할을 위해서 기존 방법과 다른 개선된 외부힘을 정의하는 것이다. 임상의 실험은 흉부 컴퓨터단층영상에서 진단에 필요로 하는 폐 실질의 분할이 성공적인 결과를 나타내었다.

• 대한의용생체공학회:의공학회지
• /
• 제44권5호
• /
• pp.303-314
• /
• 2023

This study aimed to assess the practical feasibility of advanced deformable image registration (DIR) algorithms in radiotherapy by employing two distinct datasets. The first dataset included 14 4D lung CT scans and 31 head and neck CT scans. In the 4D lung CT dataset, we employed the DIR algorithm to register organs at risk and tumors based on respiratory phases. The second dataset comprised pre-, mid-, and post-treatment CT images of the head and neck region, along with organ at risk and tumor delineations. These images underwent registration using the DIR algorithm, and Dice similarity coefficients (DSCs) were compared. In the 4D lung CT dataset, registration accuracy was evaluated for the spinal cord, lung, lung nodules, esophagus, and tumors. The average DSCs for the non-learning-based SyN and NiftyReg algorithms were 0.92±0.07 and 0.88±0.09, respectively. Deep learning methods, namely Voxelmorph, Cyclemorph, and Transmorph, achieved average DSCs of 0.90±0.07, 0.91±0.04, and 0.89±0.05, respectively. For the head and neck CT dataset, the average DSCs for SyN and NiftyReg were 0.82±0.04 and 0.79±0.05, respectively, while Voxelmorph, Cyclemorph, and Transmorph showed average DSCs of 0.80±0.08, 0.78±0.11, and 0.78±0.09, respectively. Additionally, the deep learning DIR algorithms demonstrated faster transformation times compared to other models, including commercial and conventional mathematical algorithms (Voxelmorph: 0.36 sec/images, Cyclemorph: 0.3 sec/images, Transmorph: 5.1 sec/images, SyN: 140 sec/images, NiftyReg: 40.2 sec/images). In conclusion, this study highlights the varying clinical applicability of deep learning-based DIR methods in different anatomical regions. While challenges were encountered in head and neck CT registrations, 4D lung CT registrations exhibited favorable results, indicating the potential for clinical implementation. Further research and development in DIR algorithms tailored to specific anatomical regions are warranted to improve the overall clinical utility of these methods.