The assessment of nonmetallic inclusions in steels according to the JK classification scheme of ASTM E45 using the E1122 image analysis method has presented several problems for instrument manufacturers. The major problems encountered are: discrimination of inclusion types, speed of analysis; and relating the measurement data to other methods, such as JIS G0555 and DIN 55602, as well as the new E1245 stereological method. A new technique has been developed for such ratings where the characteristics of every inclusion (type, length, width and position) are recorded in computer memory. Sulfides and oxides are first distinguished with the results stored in two different files as each field is scanned. Measurements and calculations are only performed at the end of the multiple field run over the prescribed specimen area. Advantages of this approach are numerous. Because every inclusion's position is mapped in a giant matrix, algorithms can be written and applied to separate oxide types according to their spatial distribution and shape. For instance, B-type (aluminate) stringers can be easily differentiated from C-type (silicate) stringers because B-types will generally begin and end with individual particles. Therefore, even if an agglomerated B-type stringer has the appearance of a C-type stringer, which would confuse most image analyzers (due to the continuity of touching particles within the stringer), this additional condition to the algorithm would properly classify such a stringer. Working with the stored inclusion information also allows for complete reconstruction of B- or C-type stringers that are separated by less than the 127 μm spacing (per ASTM E45 and E1122) even if the total stringer length exceeds the standard ASTM field size. Thus, “oversize” stringers that extend beyond the frame borders will be correctly sized. Another advantage of the new approach is that, regardless of the magnification used to detect the inclusions and store the data, the rating is always performed using the required 0.5 mm2 field area. Thus, this method permits used of a higher magnification which may be required to properly detect and classify inclusions in steels with low inclusion contents, such as bearing steels. Finally, because all detected inclusions are re-mapped on the screen at the end of the analysis, the inclusions can be subjected to other measurements, not required by E45 or E1122, such as those required by other standards. The end of analysis map can also be subjected to additional processing for evaluation of clustering or banding tendencies.