Microscopica acta. Supplement最新文献

The last years have seen an explosion of systems in image analysis. It is hard for the pathologist or the cytologist to make the right choice of equipment. All machines are stupid, and the only valuable thing is the human work put into it. So make your benefit of the work other people have done for you. Chose a method largely used on many systems and which has proved fertile in many domains and not only for your specific to day's application: Mathematical Morphology, to which are to be added the linear convolutions present on all machines is a strong candidate for becoming such a method. The paper illustrates a working day of an ideal system: research and diagnostic directed work during the working hours, automatic screening of cervical (or other) smears during night.

Methodological and technological requirements for cell image analysing systems are discussed with respect to the following applications: automated prescreening systems for the early detection of tumors; cell analysis systems for diagnostic assistance; early detection of benign or malignant changes in the field of "environmental pathology". The solution of these problems requires sophisticated image processing systems (high spatial resolution) which take use of supervised learning algorithms based on representative data banks. The wide spread applications in the field of automated and analytical cytology need processing systems which have the capability of systematical parameter adaptation to different analysis problems.

The different validity of DNA-cytophotometric parameters to morphometric ones for the automated cytological diagnosis of cancer is discussed and demonstrated for prostatic-, cervical- and lymph-node cytology by own measurements. Nuclear-DNA-cytophotometry used for the automation of cytological diagnosis of malignancy has the following advantages compared to image analysis: 1) DNA-cytophotometry mostly yields more sensitive parameters for the diagnosis of malignancy than image analysis does and thus results in greater diagnostic accuracy. 2) Cytological diagnosis of malignancy may often be possible in earlier stages of the tumor-history with nuclear-DNA-parameters compared to morphological parameters. 3) The prognostic validity of nuclear-DNA-distribution is probably greater than of cytomorphological parameters. 4) The parameters of nuclear-DNA-distribution for cytodiagnosis and grading of malignancy are the same for most of the human tumors in contrast to morphological parameters which are different for histogenetically different tumor. 5) Processing time for the automated diagnostic procedure of the slides will be much shorter if the Feulgen-stain for DNA is used instead of the Papanicolaou-stain for image analysis. One disadvantage of nuclear DNA-measurements as a diagnostic toll is the fact that cytopathologists cannot fully control the diagnostic process and results with their own eyes in Feulgen-stained slides. Meanwhile, however, even in other medical disciplines diagnosis are made by objective measurements of parameters which are not discernible by the human eye. That diagnostic procedure has to be accepted which yields the most accurate and reproducible diagnosis in the shortest time. It is recommended to combine cytomorphological and DNA-cytophotometric parameters as diagnostic tools for automated cytodiagnosis in Feulgen-stained slides.

This paper describes the application of a television-based system (LEYTAS) in machine analysis of cervical specimens. LEYTAS basically consists of a Leitz microscope, the texture analysis system (TAS, Leitz), a TV camera, a 4-bit grey value memory and a minicomputer (PDP 11/23). A series of 1176 Feulgen stained cervical smears has been analysed in an automated procedure using image transformation for cell selection and artefact rejection. During the analysis of the entire series new artefact rejection procedures have been added to the automated analysis. This addition decreased the percentage of false positives to 15% in the last 277 analysed smears. This percentage concerns machine performance alone. A rapid visual interaction procedure decreases this percentage to 9%. Out of 210 morphologically positive smears (diagnosis Papanicolaou class 3b or more) from the total series, one smear was missed by LEYTAS using the machine classification.

The screening by an automatic cell analyser (A) is justified, only if the false negative rate beta (A) is less than the false negative rate beta (M) of the manual method (M). A design is considered to get the numbers of ill persons who are "positive" by the manual method and "negative" by the automatic one, and vice versa "negative" by the manual and "positive" by the automatic. The sign test can be used to test the difference beta (A) - beta (M). Formulas for the necessary number n are given.

Automated prescreening means to replace visual prescreening by cytotechnologists by a machine. The aim is to reach a better diagnostic level by eliminating human failure. All machine positives whether true positive or false positives have to be examined visually (by cytotechnologists or cytopathologists). If there are many false positives the system will be highly uneconomic. From a clinical point of view the number of false negatives is crucial. The false negative rate may in no case be higher than that reached by cytotechnologists and medical cytologists, but should, if possible be lower. Yardstick for the validation of prescreening systems can only be the diagnostic level of well kept cytologic laboratories of conventional cytology. Systems which do not meet these requirements don't have the least chance of being applied in routine diagnostic cytology.