Table 5 Performance of the proposed GER detection method for the 6th channel at elementwise classification level.

From: Segmentation of gastroesophageal reflux events using a semi-U-Net architecture with 1D/2D CNNs

Episode #	Semi-U-Net model	Acc (%)	Sen (%)	PPV (%)	F1	Duration* (s)	TE* (s)
197	2D-1D	90.8167	91.4538	66.7383	0.7717	20.36	1.82
	1D	89.7000	93.6149	63.2802	0.7552		1.84
80	2D-1D	93.7667	85.5225	74.5547	0.7966	13.90	0.24
	1D	93.0750	86.5733	71.1612	0.7811		1.44
4	2D-1D	91.7778	64.8620	78.3456	0.7097	20.36	10.50
	1D	91.5278	66.8698	75.6285	0.7098		10.36
93	2D-1D	92.4125	64.6980	82.8826	0.7267	19.06	2.02
	1D	92.2417	68.3057	79.0842	0.7330		0.70
133	2D-1D	92.7083	60.0715	86.3599	0.7086	26.46	4.58
	1D	92.3806	63.7801	80.5324	0.7118		3.70
172	2D-1D	93.4762	61.9015	87.8173	0.7262	11.14	2.70
	1D	93.2691	65.7863	82.5000	0.7320		3.32
52	2D-1D	93.8625	63.9344	89.5717	0.7461	18.04	0.58
	1D	93.7562	67.8334	84.8670	0.7540		1.08
142	2D-1D	94.2019	65.9539	90.9887	0.7647	18.90	0.08
	1D	94.1500	69.6734	86.7797	0.7729		0.40
102	2D-1D	94.4750	66.2464	91.6293	0.7690	12.24	0.40
	1D	94.4183	70.0648	87.2067	0.7770		0.58
191	2D-1D	94.6606	67.5999	91.1111	0.7761	14.18	1.50
	1D	94.6545	71.5946	87.0677	0.7858		1.80
97	2D-1D	94.5944	67.7558	92.0766	0.7807	23.70	3.74
	1D	94.6972	72.0505	88.4578	0.7942		3.70
24	2D-1D	94.1141	69.2521	88.4040	0.7766	26.08	0.04
	1D	94.1923	73.4166	85.2337	0.7889		0.52
157	2D-1D	94.1893	71.7924	87.0909	0.7871	20.76	3.92
	1D	94.3643	75.6129	85.0492	0.8005		4.24
120	2D-1D	94.3956	71.7088	87.5431	0.7884	10.78	1.58
	1D	94.5944	75.6163	85.5761	0.8029		2.38
148	2D-1D	93.9896	72.1398	82.6488	0.7704	6.28	4.30
	1D	94.1573	76.1422	80.9252	0.7846		4.74
2	2D-1D	93.8951	70.9266	83.4339	0.7667	20.24	10.14
	1D	94.1265	75.1750	81.8271	0.7836		7.52
32	2D-1D	93.5500	70.6131	82.3407	0.7603	24.28	7
	1D	93.7472	74.9409	80.5372	0.7764		10.88, 4
108	2D-1D	93.6561	72.2506	83.1230	0.7731	28.12	2.72
	1D	93.8605	76.3329	81.4495	0.7881		2.52
181	2D-1D	93.4933	71.6150	83.5766	0.7713	10.42, 7	0.02, –
	1D	93.7675	76.3513	81.7716	0.7897		0.36, –
144	2D-1D	93.4683	71.8502	82.9500	0.7700	15.72	1.96
	1D	93.5063	76.2255	80.1371	0.7813		3.00
Total	2D-1D	93.5743	69.8250	84.6066	0.7603	17.52 ± 6.39	2.99 ± 3.09
	1D	93.5027	73.5129	81.3523	0.7683		3.25 ± 3.09

*Amounts are reported as mean ± std.
For the total case (averaged results across all 20 test episodes), the bold values highlight the highest scores achieved when comparing the two methods, 2D-1D and 1D.

Back to article page

Table 5 Performance of the proposed GER detection method for the 6th channel at elementwise classification level.

Search

Quick links