CASE 2242 Published on 01.10.2003

Accurate measurements using a new algorithm in a patient with hepatic metastases from colon cancer


Abdominal imaging

Case Type

Clinical Cases


Chui SL, Chui KM, Stanfield DB


76 years, female

No Area of Interest ; Imaging Technique CT, CT, CT
Clinical History
We measured the disease response to chemotherapy of a patient with known liver metastases, secondary to colon cancer, from consecutive CT examinations of the abdomen. We analysed the images using a new, patented edge definition software.
Imaging Findings
A patient with a diagnosis of primary colon cancer with metastases in the liver (confirmed at surgery) underwent treatment with Capecitabine chemotherapy. CT examinations of the abdomen and liver were obtained before and after treatment to assess tumour response to therapy.
Both pre- and post-therapy examinations were obtained using a spiral Philips CT Tomoscan AV at 120kVp, 200mA, scan time = 1 second/slice, 10mm slice-thickness. The time between the two examinations was 3 months and 4 days. Two marker lesions/metastases in the liver were selected and equivalent images on the two examinations were chosen for measurement of tumour areas for comparison. A final conclusion on the response of the disease to the treatment was then made on the basis of the change in tumour area.
The technical analysis and results are described in the `Discussion’ paragraph.

The spatial resolution of a CT scanner is restricted by the size of the X-ray tube spot and detector size. In a new, patented technique (Ref. 1), a De-Convolution Process via a Running Filter is used to pinpoint the image edge to sub-pixel accuracy (Ref. 2). A Back-Projection Algorithm transfers sub-pixels from one side of the edge to the other, helping to reconstruct the image edge for shaper conformity to the object edge, without any increase in noise (Refs. 1 & 2). Sharp definition of image-contour facilitates accurate determination of area of image profiles and their segmentation. Density within a profile may also be measured.
In this Case Study, we measured the area and mean density in CT No., on selected, equivalent slices, of two liver deposits and surrounding unaffected liver tissue before and after Capecitabine chemotherapy. We used our highly accurate image-definition software to detect the true edge of the lesion, to isolate the lesion from surrounding liver and to measure the lesion area on selected slices.


Lesion 1 on slice at -120mm: tumour area = 1941mm^2; error = +/- 0.52%; tumour tissue density [in mean CT number] = 52; standard deviation [in CT number] (S.D.) = 17.4; surrounding tissue density = 128; S.D. = 6.7.
Lesion 2 on slice at -130mm; tumour area = 2068mm^2; error = +/- 0.51%; tumour tissue density = 47; S.D. = 16.4; surrounding tissue density = 122; S.D. = 9.9.
Lesion 1 on slice at -190mm; tumour area = 3467mm^2; error = +/- 0.39%; tumour tissue density = 71; S.D. = 17.2; surrounding tissue density = 91; S.D. = 7.8.
Lesion 2 on slice at -200mm; tumour area = 2828mm^2; error = +/- 0.43%; tumour tissue density = 77; S.D. = 16.7; surrounding tissue density = 75; S.D. = 11.6.
During the examination interval, tumour areas increased from 1941mm2 to 3467mm2, and 2068mm2 to 2828mm2 respectively in the corresponding slices with the % errors listed.
During the examination interval, mean tumour densities in CT number increased from 52 to 71, and 47 to 77 respectively within the corresponding slices with their standard deviations listed.
Measurements of the mean CT number and their standard deviations for the surrounding unaffected liver parenchyma are also listed. These demonstrate a fall in mean CT density number. The scan acquisition parameters were unchanged between the two scans and the patient’s cardiac status was not noted to have changed during the inter-scan interval. The explanation for this finding is therefore likely to be altered perfusion to the liver due to increased disease in the liver hilum.


The new software allows accurate edge detection of lesions and then after isolating the lesion from surrounding tissue, allows accurate tumour areas to be calculated. This allows more accurate disease monitoring, in oncology patients who receive multiple examinations between and after courses of treatment. From tumour areas on individual slices derived from a volume scan acquisition, tumour volumes can also be calculated. The technique is not limited to tumour measurements but has a huge range of potential applications in multiple digital imaging modalities.
Differential Diagnosis List
Progressive metastatic carcinoma of the colon.
Final Diagnosis
Progressive metastatic carcinoma of the colon.
Case information
DOI: 10.1594/EURORAD/CASE.2242
ISSN: 1563-4086