# COMPUTER IMAGE ANALYSIS OF CELLS IN UROTHELIAL CANCER

> **NIH NIH R01** · MONTEFIORE MEDICAL CENTER (BRONX, NY) · 1988 · $206,170

## Abstract

Microscopic examination of cells in the sediment of voided urine is the
principal non-invasive laboratory method of diagnosis, detection and
follow-up of malignant epithelial tumors of the urothelium.  Because of
need for an objective interpretation of this difficult material, we
initiated a program of computer-based image analysis of cells in the
urinary sediment with the purpose of developing an automated diagnostic
laboratory instrument.  The potential practical significance of this
approach was enhanced further because voided urine has proven unsuitable
for analysis by flow cytometry.  In preliminary studies, using visually
selected cells, it was documented that image analysis could be reliably
used to reach diagnostic conclusions, consistent with the cytologic
diagnosis on the same material.  During the current grant period we
confirmed these results using unselected cells to reach computer-generated
diagnostic results in 40 patients with a sensitivity of 95% and specificity
of 88%.  As the next step we developed a television based, semi-automated
system of diagnostic cellimage analysis.  A trial of 119 voided urine
sediments from 103 sequential patients and 16 normal volunteers was
conducted with most encouraging results.  The performance of the system was
comparable to routine cytology for high grade tumors and was superior for
low grade tumors.  The rate of "false positive" results was high in
patients with prostatic hypertrophy and was most likely due to co-existing
urothelial abnormalities.

We propose to test these results by image analysis of 600 cytospin samples
in conjunction with cytology, histology and clinical findings.  We also
propose to accelerate image acquisition and analysis in order to create an
instrument of practical laboratory value.  The reasons why image analysis
is capable of recognizing some low grade tumors not identified by human
observers will be investigated on the premise that the cells of such tumors
may have an abnormal DNA content or other features, recognized by the high
resolution system.  We propose to continue a study of DNA content and
morphometric computer analysis of cells in urinary sediment to be compared
with DNA ploidy of tumors and clinical data.  DNA content of archival tumor
material embedded in paraffin will be measured on nuclei obtained by a
modification of Hedley's method.  Because of known significance of
aneuploidy of bladder tumors as a factor of grave prognostic significance,
this aspect of our study will also permit the assessment of the prognostic
value of image analysis of urinary sediments.

## Key facts

- **NIH application ID:** 3170312
- **Project number:** 5R01CA032345-06
- **Recipient organization:** MONTEFIORE MEDICAL CENTER (BRONX, NY)
- **SAM.gov UEI:** FP1VD1HU5HV7
- **Principal Investigator:** LEOPOLD G. KOSS
- **Activity code:** R01 (R01, R21, SBIR, etc.)
- **Funding institute:** NIH
- **Fiscal year:** 1988
- **Award amount:** $206,170
- **Award type:** 5
- **Project period:** 1982-04-01 → 1989-11-30

## Primary source

NIH RePORTER: https://reporter.nih.gov/project-details/3170312

## Citation

> US National Institutes of Health, RePORTER application 3170312, COMPUTER IMAGE ANALYSIS OF CELLS IN UROTHELIAL CANCER (5R01CA032345-06). Retrieved via MORVS 2026-10-25 from https://api.morvs.ai/grant/nih/3170312. Licensed CC0.

---

*[NIH grants dataset](/datasets/nih-grants) · CC0 1.0*