1naresh
Array
(
    [urn:ac.highwire.org:guest:identity] => Array
        (
            [runtime-id] => urn:ac.highwire.org:guest:identity
            [type] => guest
            [service-id] => ajnr-ac.highwire.org
            [access-type] => Controlled
            [privilege] => Array
                (
                    [urn:ac.highwire.org:guest:privilege] => Array
                        (
                            [runtime-id] => urn:ac.highwire.org:guest:privilege
                            [type] => privilege-set
                            [privilege-set] => GUEST
                        )

                )

            [credentials] => Array
                (
                    [method] => guest
                )

        )

)
1naresh
Array
(
    [urn:ac.highwire.org:guest:identity] => Array
        (
            [runtime-id] => urn:ac.highwire.org:guest:identity
            [type] => guest
            [service-id] => ajnr-ac.highwire.org
            [access-type] => Controlled
            [privilege] => Array
                (
                    [urn:ac.highwire.org:guest:privilege] => Array
                        (
                            [runtime-id] => urn:ac.highwire.org:guest:privilege
                            [type] => privilege-set
                            [privilege-set] => GUEST
                        )

                )

            [credentials] => Array
                (
                    [method] => guest
                )

        )

)


Table 4: 
Matrix documenting the number of unneeded rescans and recalls created by the DL approach with different thresholds, assuming that the series were scanned to rule out strokea
DL (T = 0.5)DL (T = 0.1)DL (T = 5e–4)DL (T = 1e–6)
RescansRecallsRescansRecallsRescansRecallsRescansRecalls
D125016011040
D2151833307
D32001328234
D41711036427
D52011217123
Mean ± SD19.4 ± 3.80.4 ± 0.211.8 ± 1.41.8 ± 1.37 ± 2.92 ± 1.62.2 ± 1.54.2 ± 3
  • ↵a All numbers are from the 49 test series. Here D1–D5 represent the same individuals as in Tables 1 and 2. Physician D0, whose ratings were used to train the DL algorithm, is now absent (as in Table 2) because no “stroke” ratings were available for this reader.