Timely Diagnosis of Acute Lymphoblastic Leukemia Using Artificial Intelligence-Oriented Deep Learning Methods

Table 7

Results analysis and comparison of previous works.

Reference	Features employed	Classifier (s)	Test dataset	Technical environment	Optimal accuracy (%)
Thanh et al. [13]	Morphological features	CNN (with 7 layers)	357	MATLAB	96.60
Shafique and Tehsin [14]	Color features	AlexNet (with 8 layers)	306	MATLAB	99.50
Ahmed et al. [15]	Morphological features	CNN (with 6 layers)	511	Anaconda 3 with spider 3.3 and Python 3.6	88.25
Rehman et al. [8]	Morphological features	CNN (with 7 layers)	330	MATLAB	97.78
Kasani et al. [6]	Morphological features	An aggregated CNN (-)	2100	Keras package, with tensorflow	96.58
Kasani et al. [19]	Morphological features	CNN (-)	1454	Keras package, with tensorflow	96.17
Proposed approaches	CNN features	Three various CNNs	2506	Keras package, with tensorflow	VGG-16: 97.41, ResNet-50: 95.76, proposed CNN: 85.79