SmartTracing: self-learning-based Neuron reconstruction

Brain Informatics

Table 1 The running time of each procedure and the quantitative neuron morphology measurement of 10 selected example datasets

ID	Running time (seconds)						Length		Bifurcation		Branch		Tip
ID	T _in	T _s	T _m	T _t	T _p	T _st	R _in	R _st	R _in	R _st	R _in	R _st	R _in	R _st
1	10.4	254	0.19	12.7	110	12.7	3027	4686	69	74	141	153	72	79
2	10.6	456	0.22	17.9	185	15.1	4469	7557	112	180	228	367	116	187
3	11.1	474	0.23	12.1	89	14.6	4611	6163	145	159	293	325	149	167
4	9.2	310	0.17	7.4	58	15.9	483	5823	5	117	11	240	7	124
5	10.9	310	0.19	8.5	119	16.7	3992	5635	84	92	175	188	91	96
6	9.2	29	0.17	7.5	133	22.2	176	8298	4	174	9	359	6	186
7	9.3	249	0.16	7.9	120	19.2	4408	7016	74	98	151	198	77	101
8	9.3	61	0.17	11.6	69	9.9	545	1174	7	8	14	16	8	9
9	10.1	307	0.17	9.2	53	13.4	3021	4024	75	93	155	190	81	98
10	9.0	37	0.16	7.3	78	15.3	125	3494	2	76	5	159	3	83

Visualization of the morphology of reconstructions and the original image of these examples are shown in Fig. 4
T _in generating initial reconstruction by APP2; T _s computing confidence score; T _m mRMR feature selection; T _t SVM classifier training; T _p searching foreground; T _st generating final reconstruction; R _in initial reconstruction; R _st final reconstruction; Length unit: voxel