Detection and Localization of Early-Stage Multiple Brain Tumors Using a Hybrid Technique of Patch-Based Processing, k-means Clustering and Object Counting

Table 1

Performance parameters of the proposed technique.

MRI image #	Detected pixels	Actual pixels	Precision	Recall	Specificity	Dice score coefficient	Accuracy
1	346	400	1.0000	0.8700	1.0000	0.9351	0.9996
2	472	571	1.0000	0.8266	1.0000	0.9051	0.9999
3	398	430	1.0000	0.9256	1.0000	0.9614	0.9988
4	107	110	1.0000	0.9727	1.0000	0.9862	0.9999
5	1539	1619	1.0000	0.9922	1.0000	0.9961	0.9999
6	1726	1532	0.6964	0.8051	0.9975	0.8920	0.9961
7	217	240	1.0000	0.9042	1.0000	0.9497	0.9994
8	62	63	1.0000	0.9841	1.0000	0.9920	1.0000
9	24	30	1.0000	0.8000	1.0000	0.8889	0.9997
10	510	519	1.0000	0.9827	1.0000	0.9913	0.9997
11	1859	2056	1.0000	0.9042	1.0000	0.9497	0.9993
12	2331	14	1.0000	0.9940	1.0000	0.9970	0.9999
13	4195	4320	1.0000	0.9711	1.0000	0.9853	0.9993
14	3057	3257	1.0000	0.9386	1.0000	0.9683	0.9989
15	518	582	1.0000	0.8900	1.0000	0.9418	0.9982
16	107	114	1.0000	0.9386	1.0000	0.9683	0.9999
17	509	591	1.0000	0.8613	1.0000	0.9255	0.9986
18	2869	3259	1.0000	0.8803	1.0000	0.9364	0.9885
19	1809	2161	1.0000	0.8371	1.0000	0.9113	0.9884
20	2167	2217	1.0000	0.9774	1.0000	0.9886	0.9980
Average	–	–	0.9848	0.9216	0.9999	0.9581	0.9981