4
B. Solano-Rojas et al.
neural networks (DNNs) have shown superhuman abilities in the detection of
diseases in medical computer vision, as in the work of Rajpurkar [18]. We can
design DNNs to integrate them into computer-aided diagnosis protocols for the
detection of many priority diseases. One of these possible diseases is AD.
Currently, there is a body of images of healthy patients and patients with
AD that is available through the database Alzheimer Disease Neuroimaging Initiative (ADNI)
1 . ADNI launched in 2003 as a public and private initiative. The
leadership belongs to the researcher Michael W. Weiner. The main objective of
ADNI has been to test whether medical images, other biomarkers, and clinical
and neuropsychological evaluation can be combined to measure the progress of
AD. The early detection of AD employing software would allow us to strengthen
and improve medical protocols by providing what we call Computer-Aided Diagnosis (CAD).
As we commented, DNNs have become increasingly important and useful in
recent years. One kind of these type of neural network is Convolutional Neural
Networks (CNN). CNNs are inspired by the biological visual cortex and are used
in areas as diverse as smart surveillance and monitoring, health and medicine,
sports and recreation, robotics, drones, and self-driving cars [12].
This work consists of measuring the accuracy of the detection of Alzheimer’s
disease of a three-dimensional CNN architecture, specifically a densenet-121,
trained using the ADNI MRI images. We also have a low-cost economic objective.
We aim to provide a technological artifact that has the potential of being used
in the public health and wellbeing of citizens all over the world, in particular,
for developing countries that have difficulties in accessing specialized hardware
platforms for computation.
Before presenting the results of developing a low-cost densenet for
Alzheimer’s disease detection, we first provide in Sect. 2 some background definitions to support our work. In Sect. 3 we describe previous work with more
detail. Then in the next section, we provide the methodology used to realize this
work. We present in Sect. 5 the results of the design chose. Finally, we analyze
those results with concluding remarks and future work in Sect. 6.
2 Background
We start with a short review of medical vocabulary used to provide a context
for our research. First, we introduce different clinical stages of disease that we
want to classify, and later, we describe two types of medical imaging used in the
detection and diagnosis of AD.
2.1 Clinical Disease Stages
There are different stages before the clinical diagnosis of AD. These are cognitively normal, significant memory concern, and mild cognitive impairment.
1 http://adni.loni.usc.edu.
B. Solano-Rojas et al.
neural networks (DNNs) have shown superhuman abilities in the detection of
diseases in medical computer vision, as in the work of Rajpurkar [18]. We can
design DNNs to integrate them into computer-aided diagnosis protocols for the
detection of many priority diseases. One of these possible diseases is AD.
Currently, there is a body of images of healthy patients and patients with
AD that is available through the database Alzheimer Disease Neuroimaging Initiative (ADNI)
1 . ADNI launched in 2003 as a public and private initiative. The
leadership belongs to the researcher Michael W. Weiner. The main objective of
ADNI has been to test whether medical images, other biomarkers, and clinical
and neuropsychological evaluation can be combined to measure the progress of
AD. The early detection of AD employing software would allow us to strengthen
and improve medical protocols by providing what we call Computer-Aided Diagnosis (CAD).
As we commented, DNNs have become increasingly important and useful in
recent years. One kind of these type of neural network is Convolutional Neural
Networks (CNN). CNNs are inspired by the biological visual cortex and are used
in areas as diverse as smart surveillance and monitoring, health and medicine,
sports and recreation, robotics, drones, and self-driving cars [12].
This work consists of measuring the accuracy of the detection of Alzheimer’s
disease of a three-dimensional CNN architecture, specifically a densenet-121,
trained using the ADNI MRI images. We also have a low-cost economic objective.
We aim to provide a technological artifact that has the potential of being used
in the public health and wellbeing of citizens all over the world, in particular,
for developing countries that have difficulties in accessing specialized hardware
platforms for computation.
Before presenting the results of developing a low-cost densenet for
Alzheimer’s disease detection, we first provide in Sect. 2 some background definitions to support our work. In Sect. 3 we describe previous work with more
detail. Then in the next section, we provide the methodology used to realize this
work. We present in Sect. 5 the results of the design chose. Finally, we analyze
those results with concluding remarks and future work in Sect. 6.
2 Background
We start with a short review of medical vocabulary used to provide a context
for our research. First, we introduce different clinical stages of disease that we
want to classify, and later, we describe two types of medical imaging used in the
detection and diagnosis of AD.
2.1 Clinical Disease Stages
There are different stages before the clinical diagnosis of AD. These are cognitively normal, significant memory concern, and mild cognitive impairment.
1 http://adni.loni.usc.edu.