A two-part study was designed to test the hypothesis that sufficient information is available from a modern hematology analyzer (the Coulter STKS) to reach a reliable intermediate conclusion which can be used as input to the next decision-making level in the design of a high-performance expert system for hematology diagnosis. In phase one, we analyzed the performance of three probabilistic systems (using Bayes' rule) which interpret STKS data: a control system which took the traditional approach of classifying cases into specific diagnoses, and two test systems which were designed to reach only an intermediate conclusion but not a final diagnosis. One of the test systems classified cases into "textbook categories" of disease and the other utilized defined diagnostic patterns. The systems were tested with 150 cases. The pattern approach ranked the correct choice first in 141 of 150 cases (94%). In phase two, we abandoned Bayes' rule, reformulated the pattern approach into a heuristic classification system, and tested its reliability on 820 cases. The algorithm of the reformulated system was able to classify all 820 cases into the same predominant pattern as a panel of three experienced laboratory hematologists.

Acute myeloid leukemias (AML) and myelodysplastic syndromes (MDS) enter rarely in the differential diagnosis of myelofibrosis (MF). MF of marked intensity, resulting in either “dry taps” or non-representative smears, is encountered in approximately 10% of cases. MF may be observed in any type of AML, most frequently in acute megakaryoblastic leukemia (M7). Apart from some typical cases of MDS, MF is associated with cases of acute myelodysplasia with myelofibrosis (and a major megakaryocytic component). This syndrome has been described under various headings: acute or malignant myelosclerosis, and acute MF. It should be distinguished from M7 and from myeloproliferative syndromes.

We reviewed 213 consecutive adult pancytopenic patients to determine the frequency of underlying pathology, to analyze our diagnostic procedure, and to determine the value of peripheral blood data for diagnosis. Pancytopenia was defined as the association of hemoglobin level below 12 g/dl in males and 11.5 g/dl in females, leukopenia below 4 x 10(9)/L, and thrombocytopenia below 150 x 10(9)/L. The bone marrow aspirates were normo- or hypercellular in 140 cases (66%). Bone marrow biopsies, performed in 93 cases, documented the presence of myelofibrosis in 67 cases. Aplastic anemia was diagnosed in 10% of the cases. Malignant myeloid disorders (acute myeloid leukemias, myelodysplastic syndromes, acute myeloid disorders with myelofibrosis) represented 42% of the cases and various malignant lymphoid disorders 18%. Vitamin deficiencies accounted for 7.5% and nonhematological pathology 10% of the cases. The bone marrow aspirate was sufficient for the diagnosis in 55% of the cases, and the trephine biopsy was necessary in 30%. In the remaining cases, other complementary tests were necessary to achieve final diagnosis. A discriminant analysis, focused on the hemogram data, showed that parameters obtained by analysis of blood smears were helpful for the diagnosis, especially the presence or absence of blast cells and/or of abnormal lymphoid cells.

An expert system is described that includes interpretation of the results from a complete blood count as well as data from bone marrow aspiration. The system utilizes Bayes' rule. It has previously been tested on 180 cases of anemia including 20 benign and malignant hematologic disorders. On the data set, the system achieved 84% satisfactory diagnoses. In the present study, patients with myelodysplastic syndromes and with disorders of heme synthesis have been added to the test cases. For support, the expert system requires an IBM Personal Computer or equivalent. The program is available commercially (Coulter Electronics, Hialeah, FL).

We have developed a new decision-making system which includes interpretation of complete blood count (CBC). The system works using Bayes' rule. We tested the CBC program for the diagnosis of 180 cases of anemia covering 20 benign and malignant hematological disorders. The data entered were obtained from a Coulter S + IV/HD and the interpretation of blood smears. Clinical information was not used. In 64.5% of cases, the correct diagnosis was displayed in first rank and in 20%, in second or third rank, giving a total of 84% of quite satisfactory responses. There were only 5% incorrect responses, but the proposed complementary tests rectified the error. Computer-aided diagnosis can help pathologists, clinicians, students, and technicians to make rapid correct diagnoses and choose the appropriate tests to perform. These programs run on IBM PC or similar microcomputers and are available from Coulter Electronics, Hialeah, FL.

A computer program to identify normal and pathological blood and marrow cells is presented. It requires an analysis of cellular morphology according to ten simple criteria. These criteria are treated by the Bayesian method; then the program offers, in decreasing order, the cells which are best classified. The responses are analyzed and discussed. This program seems to be well suitable to a computer-assisted teaching (CAT) of cytology. It supplements effectively the traditional methods.

The myelodysplastic syndromes (MDS) represent a group of syndromes having in common a defective production of one or more myeloid cell lines. They occur in patients which are more than 50 years old without any sex preponderance. The term MDS is replacing the obsolete and archaic term of 'preleukemia' and/or 'oligoblastic leukemia'. The more striking hematologic features are a discrepancy between a cellular bone marrow and a peripheral blood cytopenia. MDS may be idiopathic or secondary. Some of them precede or predispose to the subsequent development of an acute myeloid leukemia. A correct analysis of peripheral blood and bone marrow smears permits to classify MDS and to establish some prognostic features. Some syndromes are easily recognizable such as acquired idiopathic sideroblastic anemia, refractory anemia with excess of blasts, pure refractory cytopenia and acute myelodysplasia with myelofibrosis. Nevertheless this classification does not cover all these syndromes. Some of them with borderline features will be discussed separately. An analysis of a large series of MDS recently published in the literature will be presented as well as nosologic problems which arise. A conceptual effort should be made to recognize and evaluate the MDS.

Serum ferritin has been assayed by two methods. Radio-Immunology and Enzymo-Immunology in two group of young females. Iron deficiency is considered as unequivocal if serum ferritin is lower than 20 micrograms/l whatever the method used analysis of red cell indices in iron deficient females has shown that anemia, hypochromia and microcytosis are very late events and that their sensitivity for the diagnosis if iron deficiency is very poor.

An aid to decision programme has been applied to 200 cases of anaemia. Evaluation by the bayesian method rested on clinical data (age, sex, race) and laboratory data (blood count and differential, erythrocyte morphology). An accurate diagnosis was made initially in 107 cases, and for the first 5 diseases in 173 cases. The programme proved more effective than two clinicians recently trained in haematology. Devised for micro-computers, it can be used in routine practice and for teaching purposes.

Some patients present borderline features between acute myeloid leukemia (AML) and typical myelodysplastic syndromes (MDS): an excess of blasts insufficient to conclusively diagnose AML, yet above the figures usually compatible with MDS or the presence of Auer rods associated with a moderate excess of blasts. This presents considerable difficulties in diagnosis and management. The authors studied 28 such cases using the French-American-British Co-operative Group (FAB) classification, which groups them into a separate category termed "refractory anemia with excess of blasts in transformation" (RAEB-T). This was found to be a heterogenous group. Certain patients (4/28) had a previously established myelodysplasia, but most presented directly as RAEB-T. Two very different pictures emerged: a few patients (4/28) were young, with presentation and evolution similar to classic AML, for whom combination chemotherapy was effective; the majority (20/28) were older, with more varied clinical and cytologic presentation, for whom chemotherapy was of little effect and who presented a picture resembling classic RAEB with a median survival of 10 months.