Purpose: To compare tumor characteristics and risk factors of interval breast cancers and screen-detected breast cancers, taking mammographic density into account. Patients and Methods: Women diagnosed with invasive breast cancer from 2001 to 2008 in Stockholm, Sweden, with data on tumor characteristics (n = 4,091), risk factors, and mammographic density (n = 1,957) were included. Logistic regression was used to compare interval breast cancers with screen-detected breast cancers, overall and by highest and lowest quartiles of percent mammographic density. Results: Compared with screen-detected breast cancers, interval breast cancers in nondense breasts (≤ 20% mammographic density) were significantly more likely to exhibit lymph node involvement (odds ratio [OR], 3.55; 95% CI, 1.74 to 7.13) and to be estrogen receptor negative (OR, 4.05; 95% CI, 2.24 to 7.25), human epidermal growth factor receptor 2 positive (OR, 5.17; 95% CI, 1.64 to 17.01), progesterone receptor negative (OR, 2.63; 95% CI, 1.58 to 4.38), and triple negative (OR, 5.33; 95% CI, 1.21 to 22.46). In contrast, interval breast cancers in dense breasts (> 40.9% mammographic density) were less aggressive than interval breast cancers in nondense breasts (overall difference, P = .008) and were phenotypically more similar to screen-detected breast cancers. Risk factors differentially associated with interval breast cancer relative to screen-detected breast cancer after adjusting for age and mammographic density were family history of breast cancer (OR, 1.32; 95% CI, 1.02 to 1.70), current use of hormone replacement therapy (HRT; OR, 1.84; 95% CI, 1.38 to 2.44), and body mass index more than 25 kg/m² (OR, 0.49; 95% CI, 0.29 to 0.82). Conclusion: Interval breast cancers in women with low mammographic density have the most aggressive phenotype. The effect of HRT on interval breast cancer risk is not fully explained by mammographic density. Family history is associated with interval breast cancers, possibly indicating disparate genetic background of screen-detected breast cancers and interval breast cancers.

Extracting built-up areas from remote sensing data like Landsat 8 satellite is a challenge. We have investigated it by proposing a new index referred as built-up land features extraction index (BLFEI). The BLFEI index takes advantage of its simplicity and good separability between the four major component of urban system, namely built-up, barren, vegetation and water. The histogram overlap method and the spectral discrimination index (SDI) are used to study separability. BLFEI index uses the two bands of infrared shortwaves, the red and green bands of the visible spectrum. OLI imagery of Algiers, Algeria, was used to extract built-up areas through BLFEI and some new previously developed built-up indices used for comparison. The water areas are masked out leading to Otsu’s thresholding algorithm to automatically find the optimal value for extracting built-up land from waterless regions. BLFEI, the new index improved the separability by 25% and the accuracy by 5%.

This paper introduces a new image-based handwritten historical digit dataset named Arkiv Digital Sweden (ARDIS). The images in ARDIS dataset are extracted from 15,000 Swedish church records which were written by different priests with various handwriting styles in the nineteenth and twentieth centuries. The constructed dataset consists of three single-digit datasets and one-digit string dataset. The digit string dataset includes 10,000 samples in red–green–blue color space, whereas the other datasets contain 7600 single-digit images in different color spaces. An extensive analysis of machine learning methods on several digit datasets is carried out. Additionally, correlation between ARDIS and existing digit datasets Modified National Institute of Standards and Technology (MNIST) and US Postal Service (USPS) is investigated. Experimental results show that machine learning algorithms, including deep learning methods, provide low recognition accuracy as they face difficulties when trained on existing datasets and tested on ARDIS dataset. Accordingly, convolutional neural network trained on MNIST and USPS and tested on ARDIS provide the highest accuracies $$58.80\%$$58.80 % and $$35.44\%$$35.44 % , respectively. Consequently, the results reveal that machine learning methods trained on existing datasets can have difficulties to recognize digits effectively on our dataset which proves that ARDIS dataset has unique characteristics. This dataset is publicly available for the research community to further advance handwritten digit recognition algorithms.