TECHNICAL NOTE
Enhancement of radiographic images in patients with
lung nodules
Maher I. Rajab
1
& Ayman A. Eskandar
2
1 Department of Computer Engineering, College of Computer and Information Systems, Umm Al-Qura University, Makkah, Saudi Arabia
2 Department of Radiology, College of Medicine, Umm Al-Qura University, Makkah, Saudi Arabia
Keywords
Frequency domain processing; lung cancer;
lung nodules; X-ray enhancement.
Correspondence
Dr Maher I. Rajab, Department of Computer
Engineering, College of Computer &
Information Systems, Umm Al-Qura University,
Abdiah Campus, Makkah, Saudi Arabia.
Tel: +966 2 527 0000 ext 3021
Fax: +966 2 528 1376
Email: mirajab@uqu.edu.sa
Received: 24 February 2011;
accepted: 20 March 2011.
doi: 10.1111/j.1759-7714.2011.00045.x
Abstract
Detection of lung nodules in a chest Radiograph is very difficult due to sensitivity to
noise, lighting, and similar disturbances of the blood vessels and trachea. Therefore,
such images need to be carefully examined to identify and characterize lung lesions.
However, human interpretations are usually contradictory and may cause confu-
sion. Current works propose an image processing technique based on frequency
domain processing to clarify X-ray radiographic images taken in patients with a
variety of lung lesions. The Picture Archiving and Communication Systems work-
station allows transferring radiographic data from DICOM into JPEG image
formats. In the preprocessing phase, the lung nodules are identified by an experi-
enced chest radiologist and used for extracting regions of interest. Subsequently,
low-pass followed by emphasis high-pass frequency filters are applied to enhance the
images with appropriate cut-off frequencies. It has been found that high-frequency
domain image filtering enhances the morphological features of lung masses.
Enhanced images are then visually arbitrated by an expert radiologist.We found that
the characteristics of lung lesions are easily identified after this process.
Lung cancer remains the most common cancer-related cause
of death worldwide. Death subsequent to lung cancer alone
outnumbers deaths from breast, prostate and colon cancers
combined. According to the American Cancer Society, the
overall 5-year survival rate is less than 15%. However when
lung cancer is found in the early stages, the 5-year survival rate
increases to more than 50%. Currently, only 15% of lung
cancer is detected in the early, most treatable stages.
Chest radiography is the most frequently performed radio-
logical imaging study and also one of the most challenging.
1
Although, chest radiograph can detect early lung cancer, as a
new study from the National Cancer Institute (NCI) has illus-
trated, the false positive rate is still high. Furthermore, chest
radiograph used for detecting lung cancer is the second most
common cause for malpractice cases among radiologists as a
result of observer errors including recognition, decision
making and lesion conspicuity.
In the era of digital diagnostic radiography, denoising and
enhancement have an important potential role in obtaining
as much easily interpretable diagnostic information as
possible without increasing the radiation dose to the patient.
2
Furthermore, due to the increasing usage of high resolution
and high precision images with a limited number of human
experts, the computational efficiency of the denoising and
enhancement becomes more important.
Because of the large difference in the densities of the lung
and other structures, the chest radiograph image uses a
wide-range intensity distribution, which creates difficulty for
focussing.
3,4
In this paper we propose an algorithm for the
enhancement of chest radiographic images of lung lesions in
patients with malignant and benign lung lesions for better
detection, decision making, particularly in determining the
growth rate of the nodule, and characterization of the disease.
Image processing techniques based on frequency domain
analysis via Fourier transformation is proposed to enhance
chest radiographic images of lung lesions. Our work method-
ology is formulated so that the suggested enhancement
method applies the filtering in the frequency domain includ-
ing low-pass filtering, basic high-pass filtering and high-
frequency emphasis filtering.
5
Thoracic Cancer ISSN 1759-7706
109 Thoracic Cancer 2 (2011) 109–115 © Tianjin Lung Cancer Institute and Blackwell Publishing Asia Pty. Ltd